[SI-LIST] 20H Rule revisited
- From: "Grasso, Charles" <Charles.Grasso@xxxxxxxxxxxx>
- To: si-list@xxxxxxxxxxxxx
- Date: Mon, 2 Dec 2002 12:50:19 -0700
Hi all,
As the 20H rule seems to have resurfaced again, I thought I would resend the
full explanation from the originator
W.M King. This provides some pertinent historical background and technical
information as to the orgination of the undercut
Vcc plane concept.
Best Regards
Charles Grasso
Senior Compliance Engineer
Echostar Communications Corp.
Tel: 303-706-5467
Fax: 303-799-6222
Cell: 303-204-2974
Email: charles.grasso@xxxxxxxxxxxx; <mailto:charles.grasso@xxxxxxxxxxxx; >
Email Alternate: chasgrasso@xxxxxxxx <mailto:chasgrasso@xxxxxxxx>
*****************************************STARTS
****************************************************************************
*****************************
----------------------------------------------------------------------------
------------
Circuit Boards, Planar Impedance Reflections, Resonances, Edge Coupling, and
genesis of what came to be termed "The 20-H Rule".
Observations and discussion.
During the time frame from the mid-70's to the approximate mid-80's, several
manifestations were observed as exhibited in EMI propagation data derived
from circuit boards and their placement with respect to conductive chassis
structures.
a. The impact of distributive coupling between the circuit board and the
chassis "plane" was noted in modes of both localized eddy currents (now
approximately termed the "surface patch" effect) and as a common-mode
distributive, sequential, line. The mode of propagational operation between
these effects was observed to exhibit a causal relationship with the phasing
of current as it propagated across the circuit board planes along with the
related "image" in the chassis. Sequential current flows, as would be
experienced with a logic bus, tended to operate as a consecutive line where
unique devices (such as a single device that was asynchronous to the others
on the board) would present a localized eddy current respect to the chassis
plane surface.
b. The influence of "distributive inductance" exhibited by what I now term
"patterned layout inductance" caused by via anti-pads (holes) in the ground
and power planes was investigated, noting that these "patterns" were
presented as a phased-array in circumscription of the circuit devices. These
"array effects" significantly modified the impedance of the common-mode
distributive line that was established between the chassis plane (surface)
and the circuit board as the excitation member of the "line". The
relationship between the mode of operation as either a localized eddy
current (surface patch) or a sequential cascade of a distributive line, was
predicated upon the shape-formation of currents, and their phase
relationships, across the "patterned layout inductance". These, in turn,
coupled through fields onto the chassis plane as a potentially mixed-mode
distributed line. In this same period of time, 20 years ago, the intensity
of the "peak power currents" (now termed "Ipp" for purposes of my
descriptions) that were demanded into circuit devices and displayed in
magnitudes that were many multiples of the cumulative signal currents, were
studied and found to be sourced by a combination of the instantaneous
charging of capacitive effects in both circuit die and layout. These effects
also operated in conjunction with cross-conduction edge-transitions between
the potential rails across various drivers. (Note that the spatial field
transfer impedance between a board and a chassis plane distanced at
approximately 0.5 cm in the near-field/magnetic dominant condition, will
approximate < 40 Ohms at frequencies below 1 GHz.)
c. Resonances in the spectral patterns propagated from the circuit boards
themselves and in an inter-relationship between the transfers from the
boards to the adjacent chassis plane (or other circuit boards, if stacked or
co-planer) were observed, particularly in examples of 4 layer boards wherein
the logic ground (L0) plane was separated from the V-plane by perhaps 30 or
40 mils. In these conditions, the intense "Ipp" currents migrating as flux
between the power planes were noted to form eddys around the anti-pads, and
inducing power current edge transients into the series inductance of the
(signal) via barrels themselves. These observations sequentially yielded
other conclusions, including the inadequacy of flux-cancellation due to
phase shifts as dichotomies resultant from the via transitions as well as
the phase relationships of the flux image between the logic signals
pull-up/down currents with respect to the currents sourced in signals of
various devices. The net dynamic result was that the anti-pad array effect
because sequential inductance within the common-mode formations of the
board.
d. Concurrently, the contemplation of "Ipp" current flows between the planes
of a circuit board yielded curiosities about how these currents propagated
into the circuit devices, particularly in terms of topology. It was noted as
highly probable that when circuit devices were located near the edge
extremities of a circuit board, and that when the "Ipp" storage capacitor
was located at the periphery of the device which in turn was at or near pins
close to the edge, the flux return capture would be perhaps inefficient at
the edge which in turn probably propagated into chassis structures as fields
that were in very close proximity to the edges of the board. Indeed, studies
with local (electro-statically shielded) loop probes provided the suggestion
that in certain conditions of current paths in topology as related to the
circuit board edges, an intense fringing of the flux as a coupling
edge-field could be exhibited. Though conditionally related to the proximity
and dimension to the board-edge, with a dimensional dependency upon the
separation of the planes, the edge-field distributed transfer could result
in a coupling coefficient to other structures in the low tens of Ohms. These
transfers, assuming that the circuit board was not directly referenced to
return the flux-transferred current in "ground stitches", would impact the
potential difference between the board and the chassis plane, and hence
across the distributed line in either the localized eddy current (patch) or
distributive sequential mode depending upon the proximity of either with
respect to the edges. As a concept, the inspection of how, and where, and
in what topological distribution, does the "Ipp" current propagate became of
high interest as related to the edges of the planes.
e. With the understanding that edge-based, "Ipp" derived flux and associated
currents could be redirected back into the circuit board conceptually by
undercutting one of the power planes with respect to the other, or
"shielded" in boards with multiple parallel L0 planes through interties of
"picket fences" constructed of inter-L0 vias, the initial concept of
"undercutting planes" evolved.
f. There were two constructions subjected to evaluation with respect to
undercutting planes. The first evaluation was directly related to the
fringing of edge flux and fields as implied or stated above, with relatively
intense "Ipp" currents positioned/propagating at or on the plane edges with
the connotation that the specifics of "patterned layout inductance" could
impact the distribution and specific direction of the current flow path. In
this consideration of approach, the implementation of plane-undercuts
appeared as related to the flux and field patterns (previously identified
and mapped) analogous to those understood as surrounding traces (which I
termed "the 3-W Rule" as only a description for purposes of relating the
"envelope" of the near proximity effects associated with signal traces and
their flux and field shapes of micro-striplines). Also observed in this
conceptual frame were the effects of dichotomies in the images for signals
in the L0 planes caused the disruptions due to signal flux eddy in
circumspection of the anti-pads from via penetration of the planes. It was
considered that if flux image disruptions from anti-pads were significant at
fast edge-times (<1nS) exacerbated by an array of via/anti-pads along the
"shoulders" of the trace flux capture, then a similar effect could be
manifested at the edges of the circuit board, assuming that the "Ipp"
current and related flux pattern were propagating in extensions toward the
edges.
g. With the curiosity initiated for the first effect related in the
paragraph above, the consecutive curiosity inspired interest when the
circuit devices were positioned at distance far away from the edges of the
circuit board. In this consideration (acting in conjunction with the first
concept above) the "typical" circuit board comprised effects that approached
a distributed "lattice" wherein the "patterned layout inductance" of the
anti-pads as arrays surrounding the circuit devices contributed to a high
non-uniformity (i.e., the pattern of the anti-pads is significantly
different in density and geometry around each device) in the common-mode
distribution of fields across the board, it was interesting to note that
edge-regions of a board might represent unterminated and accumulated
distributive "stubs" in the Z-axis. When the separation between planes was
reduced to perhaps 5 mils, with a commensurately reduced power-distribution
impedance, the equivalent impedance value of the planes would be well below
10 Ohms, though this is disturbed to a significant extent by the "patterns"
of the anti-pad array inductance equivalency. Given that concept, the
contemplative concept extended, it was probable that significant edge-inward
unpopulated board regions would be not appropriately terminated as
transmission lines, and that this effect would be exacerbated when the last
devices were of low current (higher power impedance) demand when compared to
devices positioned further away from those devices (e.g., more distant from
the edges) of intense current demand (better matching and terminating the
power plane impedance as transmission lines).
(Note: The "Ipp" edge-time demand current into a processor-type device
suggests that the loading impedance can approach dynamically 0.5 Ohms
cumulatively into all power pins of the device.)
h. In both the first and second considerations noted in over-view above, it
was perceived that undercutting the one of the planes, typically the
V-plane, toward the formation of circumscribed boundary where the planes
were actually loaded by circuit devices, would either redirect (in the
concept of the first - edge flux - consideration) the flux and field pattern
back into the circuit board in a manner analogous to the flux and field
envelop of traces noted in the "3 and 10 W-Rules" reducing edge fringing, OR
(for the second consideration) actually use the boundary of the devices,
(e.g. using the devices themselves to serve as terminations for the edges of
the V-planes) reducing reflections and possible resonances established by
unterminated regions of the planes in the board topologies.
i. In the process of these descriptions, somewhere, someone, noted that in
the period where "1206" sized devices were used in conjunction with
dual-in-line packages, and that these could be near the edges, the undercut
distance would approximate 100 mils, or about 20 times the separation of 5
mils between planes. Hence the phrase "20-H Rule" was born, without
consideration of the relevance of the application or implementation. In
actuality, the purpose of the undercut was (and is) to bring the edges of
the V-plane to the limit of the boundary upon which it is terminated,
minimizing reflections caused by distributed and unterminated "stubs" in the
Z-axis in unpopulated regions of the circuit board. This, by itself, also
tends to limit the fields at the edges of the board simply because the flux
pattern and the fields are more constrained. This approach is also useful
to consider not just upon implementation of the edge-effects at the board
extremities, but for partitioning adjacent regions utilizing "moats" and
"bridges" (also terms attributed to myself) to constrain fields that may
couple into a moated region.
Given that this construction is perceived to be of extreme simplicity, even
to the point of being seen as obvious, it is with high fascination that the
rattle and intensity of discussion surrounding this concept has been
observed. The concept, as related in a patent that was granted by E. Pavlu
and J. Lockwood related to plane termination, is simply to facilitate
termination of "stubs" that are formed in the Z-axis with plane separations
of typical dimension with the added effect of containing the edge field
fringing simultaneously. In the program "EMCT: Electromagnetic
Compatibility Tutorial" (Reference: EMCT Module One, Section C, Screens
59/128 through 67/128) the schematic equivalents of this effect are
described.
Michael
****************************************************************************
****************************************************************************
***************
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
- » [SI-LIST] 20H Rule revisited
