220 IEEE TRANSACTIONS ON MAGNETICS, VOL. 32, NO.

I , JANUARY 1996

Lawrence J. Giacoletto, Life Fellow, IEEE

Abstract-Skin-effect phenomena, the reduction of a conduc-

tor effective cross-sectional area, can be evaluated analytically
in the frequency domain in terms of well-known functions for
simple conductor geometries. These results are usually approx-
imated for high-frequency limits where conduction takes place
predominantly within a thin skin depth. This approximation is
frequently seriously in error, especially for complex time vari-
ations associated with power inverters, converters, etc. For
these power applications, time-domain analyses are required.
Results useful for complex time variations are developed in
terms of series solutions that are particularly useful for com-
puter analysis of power circuits.
Frequency-domain results are also derived for rectangular
cross-sectional area conductors, as these conductors are fre-
quently used in power applications. CONTOVR,~

Fig. 1. General conductor configuration.

INTRODUCTION
HE conduction of high-frequency current along a con-
conductor as shown in Fig. 1 with isotropic properties of
ductor was first examined by Maxwell [ 11. The phe-
0 = conductivity and p = permeability are
nomenon has been subsequently examined by numerous
individuals [2]; a useful summary of developments is
given by Wheeler [ 3 ] . VxH(?, t ) = &(?, t) v * a(?,
t) = 0 (1)
Conduction develops with time as an electromagnetic
- ai?(?, t)
wave propagates from the surface toward the center of the VXE(7, t) = -p ~

at
v E@, t ) = 0.
* (2)
conductor. At dc, the entire conductor cross section is uti-
lized; at higher frequencies, only the outer skin of the Forming
conductor is effective. Accordingly, the phenomenon is
called “skin effect.” The net result is the gradual increase VxVXE(?, t ) = vv E(?, t ) - V2E(F,t )
* (3)
of the resistance and inductance of the conductor as fre-
quency increases. there is obtained
Analyses of skin-effect phenomena have been camed a@, t)
out in the frequency domain. These results are not useful V2E(7,t ) = p0 ~
(4)
at
for time-domain analysis as usually associated with com-
puter circuit analysis. In the results that follow, prior fre- Similarly,
quency-domain analyses will be reviewed and a new re- a@?, t)
sult for a conductor with rectangular cross-sectional area V2R(?,t ) = pa (5)
.
~

at
developed. Approximate time-domain results are ob-
tained for several geometries of interest. For a straight isolated cylindrical conductor, E(F, t ) has
only a - 7 component, so (4) becomes
BASICEQUATIONS
As an electromagnetic phenomenon, analysis of skin
effects starts with Maxwell’s equations. Assuming quasi-
static operation, the equations that are applicable for a

Manuscript received November 21, 1993; revised April 12, 1995.

The author is with Michigan State University, East Lansing, MI 48825
and CoRes Institute, P.O. Box 109, Okemos, MI 48805-0109.
Publisher Item Identifier S 0018-9464(96)00089-I.

0018-9464/96$05.00 O 1996 IEEE

 GIACOLETTO: FREQUENCY AND TIME-DOMAIN ANALYSIS OF SKIN EFFECTS , 221

Additional applicable equations are 2

aH(7, t )
-pup = VXJZ(7, t )
at

and appropriate boundary conditions

X/ Rg
V(t) = Ez(<, t ) Z
a(c,t ) is tangent to the C contour.
These equations need some conceptual interpretation with
respect to the end contacts to the conductor. These con-
tacts, in principle, must be ? and t dependent. Alterna-
0
tively, the conductor can be considered as a length, 2, of 0 10 20
a much longer conductor such that the details of the end f/f,
connections are no longer significant. Fig. 2. Skin-effect impedance for a thin rectangular conductor of thickness
Y, infinite and width normalized to dc resistance. Frequency normalized to
fc = p / ( 2 a p Y Z ) = 2.18347/Y:, kHz for copper conductor.
THIN-STRIP FREQUENCY-DOMAIN ANALYSIS
For a thin strip of infinite x width, thickness Y, and
length 2, equation (4) becomes where Ro = Z/(uXY) is the dc (s = 0) resistance of the
strip of width X .
(14) In the frequency domain, with s = j w ,

The Laplace transform of this equation, assuming that

Ez(y, 0) = 0 , is
where
1
w, = characteristic radian frequency = - (21)
pay2'
The solution of this equation is
For 0 = w / w , = 1 as defining the onset of skin effect,
Ez(Y, s) = A exp ( G y ) + B exp ( - G y ) . (16) z(jw,)/Ro = 1.001388 +
j0.0833. Fig. 2 indicates the
Since E,(-y, s) = E z ( y , s), B = A frequency dependency of z(j/3)/Ro = R/Ro X/Ro. For +
E z ( y , s) = 2A cosh (dpus y ) . (17)
/3 = f/f, > 20, z( jb>/Ro= Jlis
(1 + j1)/2&.

The Laplace transform of (12) with r, Y/2 can be used

-+ THIN-STRIPTIME-DOMAIN ANALYSIS
to determine V(s). Also, using the Laplace transform of The time-domain voltage, v(t), as a response to a cur-
(8) with 7 y and considering a width X of the strip,
-+

rent, i(t), can be indicated as the inverse Laplace trans-

Y12 form of (19). One method of approximating this inverse
I(s) = xu J%Y, S)dY transform is to use a series expansion of w coth(w) as
- Y/2

4Ax0 si& (6
Y/2). w c ~ t h w = l +1- w2 - - w1 4 + - w26
-~
(18) 3 45 945
-JcLos 1 2 ...
Then,
--
4725
w 8 -w l o -+
93555

= c -(2k)!
k=O
- 22kB2k
W
2k
 222 IEEE TRANSACTIONS ON MAGNETICS, VOL. 32, NO. 1, JANUARY 1996

where B2kare the Bernoulli numbers where Jo( ) and YO( ) are Bessel functions 141 of the first
and second kind of order zero and a 2 = pas. Then,
(-l)k-12(2k)! 1
B2k =
(27r)2k 1=1
7%
I (23)

Using (22) in (19), the inverse transfonn, provided i(t)

and its successive derivatives are negligible at t = 0, is
m c1 di(t)
c()i(t) - - + + wc2, -
d2i(t)

*s
-= -7j
Rn w, dt dt2
R2

+ - - -c3 d3i(t) + 3-
c4 d4i(t) + .-- -
- d[ArJ,(ar) + BrY1(ar)]
(24) RI
w : dt3 w , dt4

c2 = -1.388888 x lop3 c7 = 1.338254 x lo-" This result follows from the application of the Bessel
function relation 141
c3 = 3.306878
c4 = -8.267196 x
X c8 = -3.389680 x
c9 = 8.586062 x v(, - n)J,-.(W) = ('q
w dw [WVJV(W)] (3 1)

1
(25)
with v = n = 1 and a similar relation for Yo(w).
If the conductor is a shunt element to u(t) voltage such The application of (11) to the inside radius indicates
that u(t) rather than i(t) is known, the resulting current, that s) = 0 since the current enclosed is zero. Ac-
i(t), is cordingly, from the Laplace transform of (9)

R&(t) = dOu(t) + w,d1 -

-
du(t) d2 d2u(t)
dt
+--w f dt2
d3 d3v(t) d4 d4u(t)
+--w : dt3
+ 3-
w , dt4
+ (26)

where the first ten di values are

do = 1 d5 = -8,655503 X = [AJl(aRl) + BY~(IYRI)].(32)

d1 = -8.333333 x d6 = 8.764843 x lop7 This result follows from the application of (31) with v =
0 and n = 1, and since Jpl( ) = - J 1 ( ) and Y-l( ) =
d2 = 8.333333 X lop3 d7 = -8.885705 x lo-* -Y,( ).
d3 = -8.432539 X lop4 d8 = 9.003103 x lop9 Equations (30) and (32) can be used to determine A and
B as

CYLINDER FREQUENCY-DOMAIN ANALYSIS

For a straight hollow cylindrical conductor of inside ra- Then, with z(s) = V(s)/Z(s) = ZEz(R2, s)/I(s),

a2Ez(r,S) 1 6% ( r s) (35)
+ - z- ,pasE,(r, s).
ar2 r ar (28) where Ro = dc resistance of conductor = Z/[on(Ri -
R:)]. Introducing a characteristic radian frequency, w, =
The solution of this differential equation is
l/(paR;) (f, = 2.1834/R&,,,, kHz for copper), ,6 = w / w ,
Ez(r,s) = AJO(ar)+ BY,(ar) (29) = f If,, and y = RI/&, equation (35) in the frequency
 GIACOLETTO: FREQUENCY AND TIME-DOMAIN ANALYSIS OF SKIN EFFECTS 223

domain becomes

(36)

Fig. 3 indicates the resistive and reactive portions of

z( jo)lRo as a function of /3 for several values of y.
1 s \"

(42)
pected.
Using asymptotic expansions for the Bessel functions and then divide the two series to yield [5]
in (36), for w >> U,,

(37)

where RO2 = Z/(o.lrR$)is the dc resistance of the solid

( a = 0) conductor. Thus, as expected, the limiting imped-
ance of a hollow conductor is independent of the internal c1 di(t) c2 d2i(t) c3 d3i(t)
radius. = coi(t) +- -+ - -+ T - -
wc dt dt2 w , dt3
For w >> w,, the conduction cross-sectional area is
approximately 2 nR2Ar, where
(43)

Ar = '4
skin depth =
7

Paw
= R2
I

$
w
2 (38) This result is valid provided i(t) and its successive deriv:
atives are negligible at t = 0. In (43), the value of the
with an associated ac impedance as given by (37). first ten cicoefficients are
,Table 1 indicates a tabulation of normalized values of
z@, y)lRo for several values of /3 and for several values CO = 1 c5 = 1.469365 X 1K6
of y. These are numerical values as from Fig. 3. Note
that (37), which is usually used for evaluating skin effect,
~1 = 0.125 c6 = -9.990874 x lo-'
is not a good approximation until w >> U,. c2 = -5.208333 X c7 = 6.801447 X
SOLIDCYLINDER TIME-DOMAIN ANALYSIS
c3 = 3.255208 x lop4 CS = -4.631829 X lo-''
For a solid (y = 0) cylindrical conductor, (35) becomes
c4 = -2.170139 X c9 = 3.154634 x lo-"
(39) (44)
For the conductor as a series element, the time-domain Similarly, if the conductor is a shunt element, the time-
voltage, u(t), associated with a current, i(t) e Z(s), can domain current, i(t), associated with a voltage, u(t) e
be formally written in terms of the inverse Laplace trans-
form of (39) as
v(t) = g-l [Z(S>Z(S)]
V(s),is

R,i(t) = C-l do [ + d, (:)+d2(:r

-- _
Ro &-I
2
+- * 1
J
V(s)

dl du(t)
One method of evaluating this inverse transform is to sub- = dou(t) + --
w, dt
+--w,d2 d2u(t)
dt2
stitute the series form of the Bessel functions

+--wd3: d3u(t)
dt3
+--wd4, d4u(t)
dt4
+--d5 d5v(t)
U: dt5
+ ... (45)
224 IEEE TRANSACTIONS ON MAGNETICS, VOL. 32, NO. 1, JANUARY 1996

2 The first equations in (43) and (45), with s replaced by

jw , form valuable low-frequency frequency-domain ap-
proximations for skin effects to complement the high-fre-
quency approximation in (37).
In order to be valid, the series representation of (43) or
R/Ro
(45) must converge. A few sample calculations will serve
to indicate possible applications. Consider first a sinu-
soidal source voltage of I' peak volts and frequency,
mf, ,applied to a resistor with skin effect with dc resistance
R,. In this case, the time-domain form of i(t) is also sinu-
soidal of frequency, mfc, as i(t) = I sin(2nmfct). Thus,
1 from (43), taking into account the phase reversals with
cosine derivatives,
W W ~ / I ( ~ ~ I / R O
= RIRO = c0-m2c, + m4c4 - m6c6 -k m8Cg -
X/ Ro
(47)
g[ W ~ ) I ~ U ~ ) I ~ R O
= XIRo = mc, - m3c3 + msc5 - mIcl
0 + m 9c 9 - e - .
(48)
10 20
f/f, 1) For m = 1 using two terms,
Fig. 3 Skin-effect impedance of hollow cylindncal conductor normalized
to dc resistance R,. Frequency normalized tof, = p/(ZzpR:).
R/RO = 1 + 5.208333 x lop3 = 1.005208
compared to R/Ro = 1.005187 using (36) exact calcula-
TABLE I
IMPEDANCE FOR HOLLOW
VALUES TUBULAR
CONDUCTOR
tion.
XIRo = 0.125 - 3.255208 x = 0.124675
R=f/€, (R ,y ) /Rg a=€/€, z(R,Y) /Ro
Y = R ~ / R=zO y=Rl/R~=0.6 compared to X/Ro = 0.124676 using (36) exact calcula-
1 1.005187tj0.124676 1 1.000433tj0.041715
tion.
5 1.118044tj0.588426 5 1.010758+j0.207804 2) For m = 5 using four terms,
10 1.372087tj1.024951 10 1.042298tj0.410912
20
50
1.852574+j1.548202
2.768108tj2.476725
20
50
1.158424tj0.787444
1.693485tj1.586577 RIR, = 1 + 25(5.20833 x - 625(2.170139
100 3.798576tj3.520241 100 2.433777tj2.274441
y=R~/R2=0.2
X lop5) + 15625 (9.990874 X lo-*) = 1.118206
Y = R ~ / R z = O8.
1 1.003879+j0.111141 1 1.000032+j0.011944 compared to I . 118044 using (36) exact calculation
5 1.090879tj0.533371 5 1.000792+j0.059784
10 1.304992tj0.959143 10 1.003163tj0.119319 X/Ro = 5(0.125) - 125(3.255208 X
20 1.766451tj1.492143 20 1.012594+j0.237923
50 2.658831tj2.376933
100 3.646565tj3.379502
50
100
1.076310+j0.582754
1.275552tj1.090805 + 3125(1.469365 x lop6)
y=R1/Rz=0.4
- 78125(6.801447 x lop9) = 0.588370
1 1.001745+j0.078906
5 1.042524tj0.387993 compared to 0.588426 using (36) exact calculation.
10 1.157696tj0.739316
20 1.491836tj1.271836 3) For m = 10 using five terms,
50 2.330234+j2.093060
100 3.191170tj2.955434 RlRo = 1.357409
compared to 1.372087 using (36)
where the values of the first ten diare
XIRo 1.034947
do = 1 ds = -1.069248 x lop4
compared to 1.024951 using (36).
dl = -0.125 d6 1.848816 x lo-' These results indicate that the accuracy of (43) im-
proves as more terms are used, and that more terms are
d2 = 2.083333 x d7 = -3.196859 X lop6 required for the same accuracy as the magnitude of the
d3 = -3.580734 X lop3 ds = 5.527843 X lo-' derivatives increases. Similar calculations using (45) for
a shunt element indicate that the convergence is poorer
d4 = 6.184882 x lop4 dg = -9.558471 x lo-'. and that more terms are required for the same accuracy as
(46) with a series element.
 GIACOLETTO: FREQUENCY AND TIME-DOMAIN ANALYSIS OF SKIN EFFECTS 225

Time
(a)
Fig. 4. Skin-effect response for 1-0 conductor with f, = 10 kHz (d =
0.93455 mm) to a square wave with 5 ps rise and fall times and 20 ps
duration. (a) Current response to 1-V peak voltage source. (b) Voltage re-
sponse to a 1-amp peak current source.

The second example is for a 1-Qconductor withf, = as determined by the direction of the source current
10 kHz (d = 0.93455 mm). A 1-V peak square-wave transition.
source voltage with approximately 5 p s rise and fall times Another method of performing the inverse Laplace
and 20 p s duration is used for a shunt element, and a sim- transform of (40) is to form partial fractions of the oper-
ilar 1-amp peak source current is used for a series ele- and and inverse transform each partial fraction in turn.
ment. The computer-simulated source must have contin- Using the development of Mocanu [7], for the series ele-
uous successive derivatives [6]. Since the response wave ment, the resulting v(t) to an arbitrary source i(t) is
shape is not known, incremental computer circuit analysis
must be employed. In this analysis, the di(t)/dt term can
be modeled with an inductor, L = c 1 R O / henries.
~, The
other derivative contributions can be modeled with a com-
puter using dependent voltage sources. For the shunt ele-
ment with an applied source square-wave voltage, the
current response with contributions from the first four wherejl, are the successive zeros of Jl(z) = 0 withjl, =
terms of (45) included is shown in Fig. 4(a). Immediately 0. The following ten zeros are shown in Table I1 [8].
following the source voltage transitions, the resistance in- If i(t) = ZoU(t),then
creases abruptly to R = 6.54 n; the load current changes m

slowly and then recovers to a final value. For a series (50)

element with an applied source square-wave current, the
voltage response with contributions from the first four v(O+) = CO since, at t = O + , the element conduction
terms of (43) included is shown in Fig. 4(b). Immediately cross-sectional area is zero. Equation (50) summed from
following the source current transitions, the element i = 1 to 10 produces a v(t) that is very similar to that
voltage makes a sharp negative or positive transition shown in Fig. 4(b).
226 IEEE TRANSACTIONS ON MAGNETICS, VOL. 32, NO. I , JANUARY 1996

I
I

I
I
I
I
8 I
I
I0
I

. . . . . . . . . . . . . . . . ./ . . . . . t
VOLTFlGE # I T H SKIN EFFECT . 8

-0.1
II

I
I
I
I
II
. . . . . . . . . . . L I

, . . . -.. . . . . . . . . . . . /VOLTjeGE'WITHOUT SKIN EFFECT

c
0

o.o+ . . * . .
I

,
0
4

. . . . . . . .! . . . . . . . . . ' i
,
I

+
sous 60us 70us 80us 9ous lOOus
Time
@)

Fig. 4. (Continued)

TABLE I1
ZEROSFOR Jo(z) and J , ( z )

i jOi
1 2.40482 55577 3.83171
2 5.52007 81103 7.01559
3 8.65372 79129 10.17347
4 11.79153 44391 13.32369
5 14.93091 77086 16.47063
6 18.07106 39679 19.61586
7 21.21163 66299 22.76008
8 24.35247 15308 25.90367
9 27,49347 91320 29.04683
10 30.63460 64684 32.18968

If i(t) = ZO sin (wet) U(t). then

On a steady-state basis, the exponential term becomes

negligible and the resulting v(t) can be compared to the as in (45) is
frequency-domain results. For a summation from i = 1 to
10, the results for RIR, are not as good as those tabulated
following (48), but the results for XIR, are about equiv-
R,i(t) = 4w,
m

1 = 1
5:
e-Jo:wct eJo:wcu
v(u) du (52)

alent . wherejO, are the successive zeros of Jo(z) = 0 as shown

Similarly, the partial fraction result for a shunt element in Table 11.
GIACOLETTO: FREQUENCY AND TIME-DOMAIN ANALYSIS OF SKIN EFFECTS 221

If v(t) = VoU(t), then To satisfy boundary conditions at x = + X / 2 and y =

f Y / 2 , sequences of E,,,., ( x , y , s ) terms are needed as
(1 - e - J o f U c ' , ]
[
m

Roi(t) = 4V0 i=l jo;

At t = 0, Roi(0) = 0, and at t = CO, Roi(cn) = V,, since

U@). (53)
=
y ,m
c Ezpxn(s)cos (7)
n=l [cosh d
E] y

4ClljOj! = 1. Equation (53) summed from i = 1 to

10 produces an i(t) very similar to that shown in Fig.
4(a).
* m=
m

c Ezpym(s)
I COS (?)[cosh ,/(yr- a'x].

If v(t) = Vo sin wo tU(t), then (57)

For steady-state operation, a summation from i = 1 to 10

gives results for conductance that are not as good, and for
susceptance about equal to frequency-domain results for Fory = + Y / 2 , E,(x, + Y / 2 , s ) = V(s)/Z. Multiplying the
admittance similar to those for impedance in (47) and (48). resulting (57) by cos ( n r x l X ) , and integrating from x =
The partial fraction results cannot be easily imple- - X / 2 to x = X / 2 , produces an equation for the evaluation
mented using computer circuit analysis programs as would
of Ezpxn(S> as
be required when the driving current or voltage is not
known a priori.
RECTANGULAR CONDUCTOR
ANALYSIS
FREQUENCY-DOMAIN
sin (?)I( 7)
For a straight isolated conductor with rectangular cross-
sectional area of width X , thickness Y , and length 2,the
Laplace transform of (6) written in terms of rectangular
coordinate variables is

Similarly, multiplying (57) by cos (nrylY) and integrat-

ing from y = - Y / 2 to y = Y / 2 yields
A particular solution of this partial differential equation is

/ m a \ I/ma\

= Ezpx(s)cos (y) [coshJ(y)? - azy]

+ E,,(s) (y ) J(yr
cos cosh - a 2 x ] (56)

where, as before, a2 = - p m . Thus,

sin (7) [cosh J(;r - a 2y ]

m
228 IEEE TRANSACTIONS ON MAGNETICS, VOL. 32, NO. 1 , JANUARY 1996

The s-domain conductor current is

Y12

c
l m
+ -Xm=1
j(q - (Y 2

where R, = Z/(oXY) is the dc (s = 0) resistance of the conductor. Forming V(s)/Z(s),the s-domain impedance of the
conductor is

In the frequency domain with s = j w , and introducing a characteristic radian frequency, w, = 4/(0pXY),

For w = 0 and X/Y >> 1, z(0) = R, since tanh x / x + 1 = 1 mm; f, = 436.694301 Hz for copper, z(jw,)/Ro=
as x + 0 and C(2n - 1)-2 = n2/8; similarly, for w = 0 1.001674 + j0.016191, and z(j20wC) = 1.024015 +
and Y/X >> 1, z(0) = R,. For X = Y = 1 mm using 20 j0.321481. Next, consider using about the same mass of
terms in the series, z(0)/Ro = 1.000161; with fc = copper to form a hollow tubing of 1-mm wall thickness
8.733886 kHz as for copper, z ( j w c ) / R o = 1.007588 + and R2 = 20/(2a) = 3.1831-mm outside radius; f, =
+
j0.14003, and z(j20w,)/Ro = 1.933488 j1.54761. For 215.50 Hz, z(jw,)/R, = 1.000177 + j0.027378, and
comparison, a solid copper cylindrical conductor with z(j20w,)/Ro = 1.068476 + j0.534892. From this limited
R2 = 0.5 mm (inscribed circle), f, = 8.733886 kHz; comparison, it would appear that, on a mass basis, a hol-
z(jw,)lR, = 1.005187 + j0.124676, and z(j20w,)/R0 = low tubular conductor is more effective than a thin rect-
1.852574 + j1.548202. For another comparison, con- angular conductor at low frequencies, but less effective at
sider a thin rectangular conductor with X = 20 mm and Y higher frequencies.
 GIACOLETTO: FREQUENCY AND TIME-DOMAIN ANALYSIS OF SKIN EFFECTS 229

material and given X , Y values can be obtained using com-

2 puter formulations. Similarly, frequency-domain [91-[ 111
1 and time-domain [lo], [l11 results for conductors with
arbitrary geometries can be obtained using finite element
computer software.

1 CONCLUSIONS
A method of introducing skin effects in time-domain
circuit analyses has been developed. The significance of
skin effects can thus be determined in circuits with arbi-
trary signal variations as, for instance, in pulse switching
circuits. A new result has been developed for the skin- *
effect impedance of a conductor with arbitrary rectangular
cross-sectional area.

ACKNOWLEDGMENT
The author would like to thank a reviewer for correc-
tions and comments as evidenced by careful reading of
the text.

REFERENCES
[1] J. C. Maxwell, A Treatise on Electricity and Magnetism, vol. 2, sec.
690, p. 385, 1st ed. 1873, 3rd ed. 1892 (reprinted 1904, 1937 p. 322,
20 1946, 1955), London: Oxford University Press.
[2] K. Simonyi, Foundations of Electrical Engzneen‘ng (trans. D. Jones
and A . Sekey). New York: Macmillan, 1963, pp. 494-510.
[3] H. A. Wheeler, “Formulas for the skin effect,” Proc. IRE, vol. 30,
Fig. 5. Skin-effect impedance of rectangular conductor normalized to dc pp. 412-424, Sept. 1942.
resistance Ro. Frequency normalized to fc = 2p/(7rpXY). [4] M. Abramowitz and I. Stegun, Eds., Handbook of Mathematical
Functions. New York: McGraw-Hill, 1941, pp. 358-494; repub-
lished by Dover, 1961.
TABLE 111 [5] L. J. Giacoletto, Ed., Electronics Designers’ Handbook. New York:
IMPEDANCE
VALUES
FOR RECTANGULAR CROSS-SECTION
CONDUCTOR McGraw-Hill, 1977, pt. 1-23.
[6] L. J. Giacoletto, “Altemate PSPICE switch functions,” IEEE Cir-
flf, z ( f / f c ,X/Y)/RO f/fc z ( f/f,,XlY 1 IRo cuits and Devices, vol. 7, pp. 9-37, Sept. 1991.
[7] C. I. Mocanu, “The equivalent schemes of cylindrical conductors at
X/Y=1 X/Y=7 transient skin effect,” IEEE Trans. Power Apparatus and Syst., vol.
1 1.007588tj0.140030 7 1.023305tj0.300937 91, pp. 844-852, May-June 1972.
5 1.160391tj0.640051 35 1.427103tj1.285045 [SI M. Abramowitz and I. Stegun, Eds., Handbook of Mathematical
10 1.457128tj1.059625 70 2.055274tj1.984348 Functions. New York: McGraw-Hill, 1941, p. 409.
20 1.933488tj1.547610 140 2.919650+32.780555
50 2.838494tj2.479170 350 4.519784tj4.394392
[9] A. Konrad, “Integrodifferential finite element formulation of two-
100 3.868554tj3.524242 700 6.338753tj6.225216 dimensional steady-state skin effect problems,” IEEE Trans. Magn.,
vol. MAG-18, Jan. 1982.
X/Y=2 XlY.10 [lo] 0. Biro and K. Preis, “F~niteelement analysis of 3-D eddy cur-
2 1.018055tj0.226544 rents,” IEEE Trans. Magn., pp. 418-423, vol. 26, Mar. 1990.
10 1.023486+j0.310245
10 1.325233tj0.953455 50 1.435774tj1.335026 1111 3. R. Brauer and F. Hirtenfelder, “Surface integrals on 3D and 2D
20 1.776280tj1.472257 100 2.091230ti2.067685 finite element models for skin effect excitations and open bound-
40 2.414544tj2.092826 200 2.993739tj2.895159 anes,” IEEE Trans. Magn., pp. 1659-1662, vol. 28, Mar. 1992.
100 3.628475tj3.325053 500 4.659754tj4.574211
200 5.006249tj4.714502 1000 6.554641tj6.477658
X/Y=5 XIY-20
Lawrence J. Giacoletto (S’37-M’44-SM’48-F’58-LF’82) received the
5 1.022984tj0.288702 20 1.024015+j0.321481 B.S.E.E. degree from Rose-Hulman Institute of Technology, the M.S. de-
25 1.413924tj1.222512 100 1.444829tj1.398279 gree in physics from the State University of Iowa, and the Ph.D. degree in
50 2.007975tj1.882776 200 2.135246tj2.176113
100 2.826524tj2.642279
electrical engineering from the University of Michigan.
400 3.089318tJ3.044419
250 4.348585tj4.177634 1000 4.842076tj4.803829 He spent 10 years as a Research Engineer with RCA Laboratories. He
500 6.076976tj5.918516 2000 6.832813+j6.797039 then became Research Manager at Ford Motor Co. Scientific Laboratory
for 5 years. In 1961 he joined the Electrical Engineering Department of
Michigan State University, and became Professor Emeritus in 1987. Among
his 22 patents is the first patent for a transistorized auto ignition system
Fig. 5 indicates a series of z( jP)IR, = R/Ro + XIR, and a basic patent for UHF varactors. As a consultant with CoRes Institute,
he 1s currently extensively involved with problems related to power elec-
curves forfY/fcX = 0 to 20 and for X/Y = 1, 2, 5, 10 tronics, especially the simulation of power applications. He has edited
and 20. Table I11 shows the specific numerical values for Electronics Designers’ Handbook, authored Differential Amplijiers, and
Fig. 5. contributed to three other books.
Dr. Giacoletto has been active in IEEE activities, including serving as
Analytical results for time-domain operation become Chairman of three sections, Chairman of Fellow Awards and Field Awards
prohibitively complex. Numerical results for a specific Committees, and Member of the Board of Directors.