Page 1
Prediction and Verification of Power/Ground Plane Edge Radiation
Excited by ThroughHole Signal Via
Based on Balanced TLM and Via Coupling Model.
Jun So Pak, Junwoo Lee, Hyungsoo Kim, and Joungho Kim
Terahem Interconnection and Package Labxatory, Department of Electrical Engineaing and Computer
Science. Korea Advanced Institute of Science and Technology, 3731 Kusong, Yusong, Daejeon 305701, Korea.
Tel) +82428695458, Fax) +82428698058
Email) cliitoonebminfo.kaist.ac.kr. flusu~’~~itlro.knist.ac.kr,
and jouneho/&ee.knist.ac.kr
ABSTRACT  We introduce a modeling and simulation method to predict powedgmund plane
resonance and edge radiation coupled fmm.broken return current path of throughhole signal via. and
analyze the coupling and radiation mechanism. The approach is successfully verified with a series of
measurements with various plane conditions.
sbaniino‘~ccinfo.k~sl.~,~.kr,
I. INTRODUCTION
In recent high paformence digital systems, the clock frequency and its harmonics are continoously increased
over GHt range. At the m e time, the system density becomes higher, and then PCB and package have more
number of plane layers and more embedded signal traces, containing more thmughhole signal via and stripline.
Since these trends makes the PCB size be almost equal to wavelength of opting frequencies and, hence, the
radiated emission problem is more complicated and increased The radiakd anision problem of the high
performance digital system is too serious and difticult to be solved. Nevertheless, the radiated emission from
transmission line has been well and detailed analyzed so far. On the other band, the study on the pwer/ground
plane edge radiation problem has just started and expimentally investigated in a basic level approach. Since the
power/ground plane edge radiation has the many excitation sources, which are a large current swing U 0 driver
[l] well known as SSN (simultaneous switchmg noise) a n d the anbedded signal lrace in power/ground plane
([3] [4] [SI), the solving of the radiated emission is wholly depending on the settlement of coupling mechanism
between powerlground plane and the excitation sources.
However, these previous investigation overlooked the role of throughhole signal via, which are used for
layer transition of signal traces. The [3] assumed the &tion
free space [Z]. However the power/gmund plane is not transmission structure but cavity, which bas no travelmg
mves but standing waves. Therefore the exciting source of the pwer/gmund plane resonance and edge
radiation is not the current of throughhole signal via itself but the broken r e 
between different layas around the throughhole stgal via [6].
In this paper, we infxoduce a new modeling and simulation method to predict the pwer/ground plane
resonance and edge radiation based abalancBd TLM model and a detailed coupling via model to excite the plane
resonance. Based on the approach, we analyze the coupling mech” between the throughhole signal via and
the power/ground plane. Also we relate it to the relation between the edge radiation of the power/ground plane
and the rewnance of it And the truth that the origiaal source of the pwer/ground plane edge radiation is the
broken r e m current of throughhole signal via and verified With the model and the experimenfs. We have
successfully predicted the powedground plane edge radiation and it is well verified with a series of the
measurements with various power/ground plane conditions.
pattern of throughhole signal via is same as open
current, which has to jump
1 1 . Model and Simulation of Power/Ground Plane Edge Radiation
Excited by Throughhole Signal Via.
Firstly, we assume that the power/pmd plane has no AC current path like as &coupling capacitor, and the
top, bottom layer microstrip lines which are connected with thmoghhole signal via, have the ground plane,
power plane as return current path, respectively (Fig. 2). Therefore, when the input signal sent along with a
07803812&9/03/$17.00 0 2003 lEEE
181
Page 2
single metal l i e to termination comes back, it has to jump from power plane to ground plane around through
hole signal via. Namely, the retum current meets the barria called power/ground plane impedance. As well
known, the power/ground plane impedance is changed with position and frequency, because of power/ground
plane resonance. If the power/ground plane impedance is large at thoughhole signal via, the return current of
thoughhole signal via is difficult to go Gom power to ground plane. In other view point, this situation is
analyzed like that the voltage drop is induced between power and ground plane by the return current On the
other hand, if the powedground plane impedance is low, the induced voltage is mall This analysis is well
explained with proposed model (Fig. I, Fig. 3). The proposed model is formed with the balanced TLM, which
well represents the discontinuity retum current path in power/gmund plane, and Yia coupling model. As
mentioned previous, since the power/ground plane is the open boundary cavity, the induced voltage at through
hole signal via is appeared at the power/ground plane open edge. This open edge voltage is the magnetic current
source of the powedgound plane edge radiation ( [ 7 ] , Fig. 4).
Since the power/ground plane edge radiation is occurred by above mechanism, the radiation pattem can be
easily predicted with the Sparameter measurement of transmission line including thoughhole signal via. The
S21 magnitude value shows the loss occurred in the signal trace, besides, the edge radiation begis at the voltage
drop in power/ground plane impedance. Therefore, the additional loss of microstrip line including throughhole
signal via compared to line without via is the transferred power to power/ground plane (Fig. 5). As shown in Fig.
8 (a), we can predict that the shaded a m in Fig. 5 w i l l be expressed as a pwer/ground plane edge radiation.
111. Experimental Verification of the Model and Simulation
The proposed prediction method was verified with the measurements of powedground plane edge radiation
The radiation was picked up with the lab made loop antenna Since the magnetic cnrrent source is made in the
longitudinal edge direction (Fig. 4), the magnetic field has the snme direction. Therefore the loop antenna is
located in vertical direction, and this direction makes the radiation of microstrip l i e be cancelled. The five
DUTs were made to have&e position variation (center, offset on powedground plane), the plane impedance
variation at the same position(??a, short four edges without, with 5mm spacing decoupling capacitors m y ) ,
and no throughhole signal via line *,a refaence of the edge radiation Fig. 7 and Fig. 8 show the measurement
results of the center and offset position; respectively. And (a) and (b) are without and with the decoupling
capacitors m y , respectively. All results can be well predicted with the additional insertion loss of tmnsmission
line, which is occurred by the broken retnm current of throughhole signal via and w e l l correhtdwith
simulation.
Fig. 6 shows the reference of pwe/ground plane edge radiation, and that the environment has no otha Em
source. This result is same as no source case, because we eliminste the radiation of microstrip l i e with the loop
antenna direction. The Fig. 7 and Fig. 8 show the relation betwm the additional insertion loss and the
powedground plane edge radiation including throughbole signal via In Fig. 7 , the teasun of dflerence between
(a) and (b) is the boundary condition Fig. 7 (a) has both even number modes, and (b) is both odd. In Fig. 8,
mostly modes are shown. Only Fig. 8 (a) can not have the including ‘2’ modes, and @) can not have the
including ‘4’ modes, because the offset means one fourth position of plane width. These facts say that the
powedground plane edge radiation depends on the position of throughhole signal via and the boundary
condition, which change the powa/ground plane impedance profile, also that the decoupling capacitors m a y
can reduce the edge radiation
IV. Conclusion
In this papa; we successfully defmed that the important original source of powedpund plane edge radiation
is the broken r e m current of throughhole signal via, which induces the voltage at powerlground plane
impedance, and verified our def~tion with the simulation with the proposed model. Also, we could successfully
predict the powedpund edge radiation with ow model and simulation, and verified ow prediction with a series
of the measurements with various power/gmund plane conditions. The powedground edge radiation could also
be predicted in proportion with the additional insertion loss of signal trace.
182
Page 3
V. REFERENCES
[I 1 Sergiu Rad& David Hockanson, “An investigation of PCB radiated emissions from simultaneous switclung
noise”, L E = EMC Symposium pp. 893898,1999.
[2] Robert G Kaires, “Radiated Emissions from printed circuit b a r d traces including the effect of vias, as a
function of source, termination and board characteristics”, IEEE EMC Symposium. pp. 872877, I998
[3] Fram Gisin. Dr. Zorica PanticTanner, “Radiation fmm printed circuit board edge structures”, I E E E EMC
Symposium. pp. 881883,Aug. 2001.
(41 TM
Harads, Wi
Sa%&, TOM& Kuriyama, ‘‘Radiated emission f“ a multilaya PCB with traces
placed between power/ground planes ” , IEEE EMC symposium. pp. 253257,2002,
[SI Sabru Hag4 K e n N h o , Osamu Hashimob, ‘%eduction in radiated emission by symebical power
ground layer stack up FCB w i t h no open edge’: IEEE EMC Symposium pp. 262267,2002,
[6] Jun So Pak, Jingook Kim Heejae Lee, JungGun Byun, Joungho Kim, “Coupling of thoughhole signal via to
powa/gmund rsnnance and excitation of edge radiation in multilayer PCB ,2003 IEEE EMC Symposium.
will be published in Aug. 2003.
[7l ConstantineA. Balanis, “AnkmaTheory”, second edition, John wiley & wns, hc.,
1997.
~0~ ay er M i  p
~ i m ThnwghHole SigMl Via
~
Balmcad RM
Figure 1. The proposed model of power/gmund plane including throughhole signal via. The
power/gmnnd plane is modeled with balanced TLM, which well represents the physical
struchre. Miemstrip line bas no coupling to powerlgmuud plane except for thmughhole signal
via.
Ground
Reblm Oment Pam
Power
. .
FigureZ The signal path and return current path of the micmstrip line including the thmughhole signal
via The return current of throughhole sigual via has to jump fmm a gmund plane to a power
plane. So the return current induces the voltage, which is voltage dmp by pnwer/gmund plane
impedance.
F~ure
3. The pmposed model is verified with
the measurement. Model can .well
predict the powedgmund
resonances and the increased insertion
loss of the microstrip lines hy the
bmkeu return current of the thmugh
hole signal via
plane
0.6
1.
183
Page 4
F ~ u r e
4. The mechanism of the powerlgmund plane edge radiation. The voltage shown at open edge is
calculated f r o m the induced voltage at throughhole signalvia
FreWlElq
(W
Figure 5. The increased insertion loss due to the
thmughhole signal via is used to predict
power/gmund plane edge radiation like
as Rig. 8 (a). (Measurement)
. ThouihHole Simd V i : Cenber
. DeCoupling Capacibx : M
E l I l r B c m d R n s t n u n n r a d s U o * w m ~ ~ P ~
1
U~~~UIC
rii~mrolmtme
S ~ W M
1s
8
s
FreQnW
IWr)
Figure 6. The nefemnce of the powedgmund
plane edge radiation. The throughhole
signal via does
powerlgmund resonance and plane edge
radiation.
. Thouphilols Sipd V m : Cenfer
not excite the
B
#
I I;

%:sJ
7
'0 ic
a0
a
f

IWZ)
@)
0 1 1
1 1 1
1.D
1.5
IO
LI
RWem
(a)
Figure 7. The model and predictlon of the edge radiation is verified with the measurements of the center
located thmughhole signal via DIJT (a) and @) are without and with decoupling capaciton
array, which can reduce the edge radiation.
(a)
@)
Figure 8. The model and prediction of the edge radiatlon is verified with the mearurements of offset
located throughhole signal via. (a) and @)are without and with demupliig capaciton array.
184