Progress In Electromagnetics Research C, Vol.

51, 111–119, 2014

Analytical Computation of AC Resistance of Single-Layer

Air-Core Helmholtz Coils

Weijie Luo* , Quan Ke, Guozheng Yan, and Kai Yang

Abstract—An analytical expression of AC resistance of a single-layer air-core Helmholtz coil is

presented. Proximity effects between both bundles and strands of litz-wire as well as skin effect are
considered. To obtain an accurate analytical expression of proximity effect on bundle level and strand
level, a precise distribution of magnetic field is discussed. The analytical expression of AC resistance and
quality factor is verified with the experimental results, and the theoretical predictions are in agreement
with the measured results.

1. INTRODUCTION

Power supply has been a major issue for micro medical devices that cannot use a cable to obtain energy.
In this case, battery is a possible solution. However, batteries’ limited storage capacity will not allow
medical devices to work at a comparatively high power for a long period of time. For instance, most
commercial wireless capsular endoscopic (WCE) devices used in gastrointestinal (GI) tract use batteries
as their power supply. Latest WCE devices, which are capable of taking 2 or 3 frames of GI tract image
per second and transmitting them to a receiver, can work for about 8–12 hours. Those batteries can
provide an average of 25 mW of power, which is insufficient for capsule endoscopy that requires more
power — capsule robots. Many efforts have been devoted to the development of a continuous power
supply system [1], among which wireless power transfer (WPT) [2–5] is a promising option. This system
can provide a steady and sufficient supply of energy to an endoscopy capsule.
WPT system consists of three components: (1) a driving circuit which can provide a square wave
current; (2) a power transmitting coil; (3) a power receiving coil with a rectifier and filter circuit. In
this wireless power transmission system for WCE devices, the diameter of the transmitting coil should
be over 400 mm, which is bigger than most people’s torsos, and that of the receiving coil should be no
more than 15 mm, which is the minimum diameter of GI tract. Due to the great disparity between
transmitting and receiving coil, a loosely coupled transformer with a large air gap in between is formed.
The receiving coil affects the transmitting coil less as the diameter of the transmitting coil increases.
In the meanwhile, the winding resistance of the transmitting coils becomes more significant. Thus it is
necessary to analyze winding losses in transmitting coils at high frequency. The winding AC power loss
of the coil at high frequency is caused by eddy currents, due to skin effect and proximity effect. Both
effects will become more dominated as the frequency increases.
Litz-wire is commonly used to reduce winding power losses of inductors and transformers. Since
litz-wire is multi-stranded, the overall current is distributed into each strand, which will result in a
more uniform current distribution across the wire cross section. Besides, litz-wire is also twisted so that
every strand can cover all the positions in the wire cross section along the longitudinal axis, and this
property will make sure that current in each strand is the same. Therefore, not only the skin effect
Received 28 April 2014, Accepted 18 June 2014, Scheduled 23 June 2014
* Corresponding author: Weijie Luo ([email protected]).
The authors are with the Department of Instrument Science and Engineering, School of Electronic Information and Electrical
Engineering, Shanghai Jiao Tong University, Shanghai 200240, China.
112 Luo et al.

but also proximity effect is reduced [6]. In the following analysis, it is assumed that current density
distribution is absolutely uniform in every strand.
Many approaches to analytical computation of AC resistance of multi-strand and litz-wire windings
have been presented [6–11]. However, most of the analytical modelsare developed to calculate AC
resistance of windings of inductors and transformers. Therefore these models have been derived under
the following assumptions:
1) The core material has a high magnetic permeability.
2) The winding layer wires fill the entire breadth of the bobbin.
Under these two assumptions, the behavior of the winding is close to that of an ideally infinite long
solenoid winding. But when analyzing the single-layer air-core Helmholtz coil that is used as power
transmitting coil, those assumptions is invalid, for this coil does not have a high-permeability core and
there is a huge air gap in the middle of the winding.
The purpose of this paper is to develop an analytical expression of AC resistance of a single-layer
air-core Helmholtz coil. In the analysis, the magnetic field distribution across every conductor and
every strand is considered. The influence of twisting of the litz-wire is also taken into account. The
theoretical prediction is verified with experimental results.

Notation:
δ Skin depth.
dc Diameter of a conductor.
d0 Diameter of a strand.
H Amplitude of the magnetic field.
f Frequency.
N0 Number of strands in a conductor.
NT Number of turns.
Rac AC resistance of the winding
ρ Copper resistivity.
ζ Normalized value of the diameter of a strand (d0 /δ).
I Amplitude of current in conductors.
I0 Amplitude of the current in strands.
p Pitch of the twisting.
D Diameter of the coil.
Dg Length of air gap in the middle of the Helmholtz coil.

2. ANALYSIS OF THE MAGNETIC FIELD DISTRIBUTION ACROSS A LITZ-WIRE

WINDING SPACE

Figure 1 shows the Helmholtz coil discussed in this paper. It consists of two identical solenoid coils
with NT /2 turns (NT is even). The distance between these two solenoid coils is Dg + N2T ×dc = D/2.
Because the diameter of both solenoid coils R is much greater than the length of them dc N T /2, each
solenoid coil should be considered as a row of NT /2 ring coils in order to analyze the magnetic field.
The external magnetic field across each turn (Hext ) is the accumulation of magnetic field generated
by all other turns. Each turn is considered as a current ring, thus external magnetic field distribution
across each turn (Hext ) can be derived by using Biot-Savart’s law:
1 Id l × ( r 0 − R) IR h i
* * * * *
* 1 Id l × r p * * *
dH= = = z cos θ i + z sin θ j +(R − x cos θ) k dθ (1)
4π rp3 4π rp3 4πrp3
Due to the symmetry of the ring current, as shown in Fig. 2, magnetic field at P (x, 0, z) can represent
magnetic field at any point in the space and magnetic field along y direction is 0 at P (Hy = 0). The
Progress In Electromagnetics Research C, Vol. 51, 2014 113

Nt/2

D/2
Dg

Nt/2+1

Nt

Figure 1. Single-layer air-core Helmholtz coil.

magnetic field at P is given by

Z Z 2π
IR
Hx (x, z) = dHx = z cos θdθ (2)
0 4πrp3 (x, z)
Z Z 2π
IR
Hz (x, z) = dHz = (R − x cos θ)dθ (3)
0 4πrp3 (x, z)
√
where rp (x, z) = x2 + z 2 + R2 − 2Rx cos θ and R is the radius of the coil.
Therefore, the magnetic field generated by the jth turn at the (r, ϕ) position of the ith turn is:
Hxij (r, ϕ) = Hx (R + r sin ϕ, zij + r cos ϕ) (4)
Hzij (r, ϕ) = Hz (R+r sin ϕ, zij + r cos ϕ) (5)

 (i − j) dc , i ≤ N2T , j ≤ N2T or i > N2T , j > N2T
where zij = (i − j) dc − Dg , i ≤ N2T , j > N2T .

(i − j) dc + Dg , i > N2T , j ≤ N2T
The total magnetic field in (r, ϕ) position of the ith turn is:
XNT
Hxi (r, ϕ) = Hxij (r, ϕ) (6)
j=1,j6=i
XNT
Hzi (r, ϕ) = Hzij (r, ϕ) (7)
j=1,j6=i

Thus the amplitude of magnetic field at (r, ϕ) is:

q
Hi (rϕ) = Hxi 2 (r, ϕ) + H 2 (r, ϕ) (8)
zi

The mean amplitude of external magnetic field across the ith turn is:
Z 2π Z rc
1
Hexti = 2 Hi (r, ϕ)rdrdϕ (9)
πrc 0 0
Computation of the results in Fig. 3 is achieved by using numerical calculation functions of Matlab.
Unlike magnetic field of infinite long solenoid coil, magnetic field across each turn differs greatly with
that of other turns. As shown in Fig. 3, amplitude of magnetic field is the lowest at 14, 15 turns and
42, 43 turns, which are in the middle of two identical solenoid coils, and reaches the highest points at
turns on the edges of the Helmholtz coil.
To simplify the calculation of AC resistance in the following section, the external magnetic field at
each turn is considered as a constant field with the amplitude of Hexti .
114 Luo et al.

X
Hx
Hz
rp P (x,0,z)
Idl r0 r ϕ
θ
R
0 Z
x
y
j th i th z
Z ij
(a) (b)

Figure 2. (a) Magnetic field generated by the jth current ring. (b) Cross section of the ith ring (in
xz-plane).

Figure 3. Calculated amplitude of magnetic field at each turn when I = 1 and NT = 56.

Besides the external magnetic field in the winding (H ext ), each strand in a given conductor is also
submitted to an internal magnetic field (H int ) that is generated by current in other strands in the same
conductor, as shown in Fig. 4. Since all strands in this conductor is twisted, it is assumed that each
strand has the same current, which is I/N0 .
Current vector of a single strand travels through a helix along the longitudinal axis of the conductor,
as shown in Fig. 5. Considering that current vectors in all strands have similar paths, the internal
magnetic field generated by them is not perpendicular to the cross section plane of this given conductor
nor in the plane. The internal magnetic field can be divided into two components: one is the magnetic
field perpendicular to the plane (H⊥ ) and another one is the magnetic field in the cross section plane
(Hp ). In the following analysis, we suppose that current density at any position across the conductor
has the same amplitude and slant angle.
As shown in Fig. 4, under the assumption that the current density distribution is uniform, the
magnetic field in the cross section plane (Hp ) can be given by
I
Hp (r) = r (10)
2πrc2
As shown in Fig. 5, current density J can be divided into two components: current density in the
cross section plane (Jp ) and current density perpendicular to the plane (J⊥ ). J⊥ points out of the front
of the diagram, and Jp follows the counterclockwise direction, which is also the twisting direction of
litz-wire.
Progress In Electromagnetics Research C, Vol. 51, 2014 115

Twisting
direction

Hp Jp
θ
r dr/tan θ
θ
0 J
Jp J
θ
Jp J
dr Jp

dr
(a) (b) (c)

Figure 4. Cross sec- Figure 5. (a) Two kinds of current density in a conductor: Jp and J⊥ .
tion of a conductor. (b) Geometric relationship between Jp and J⊥ . (c) Cross section of different
directions in a given strand.

The current density perpendicular to the plane J⊥ is given by

I
J⊥ = (11)
2πrc2
J⊥ and Jp is submitted to the geometrical relationship in Fig. 5(b):
Jp = J⊥ tan θ (12)
where θ is the slant angle, tan θ = 2πrp , and p the pitch of the twisting.
c

Within the shaded ring area in Fig. 5(a), Jp is continuous. Therefore, this shaded ring is equivalent
to a current ring. Since there are countless similar current rings with the same radius along the
longitudinal axis of the conductor, the magnetic field generated by these current rings is identical
to that of an ideally infinite long solenoid coil. The curvature of the wire is neglected because dc ¿ D.
Thus to obtain the magnetic field generated by Jp , which is H⊥ , we need the density of Jp along
the longitudinal axis, as shown in Fig. 5(c):
tan θ
n= (13)
dr
The magnetic field generated by Jp at r = R (R < rc ) is
Z rc
dr I
H⊥ (R) = nJp × dr × = 2
tan θ(rc − R) (14)
R tan θ 2πrc
Apparently, the direction of H⊥ is parallel to the longitudinal axis.
The total internal magnetic field is
q s
I2 I2
Hint (r) = Hp (r)2 + H⊥ (r)2 = r 2+ tan2 θ(rc − r)2 (15)
(2πrc2 )2 (2πrc2 )2

3. AC RESISTANCE OF ROUND LITZ-WIRE WINDINGS

Eddy current losses per unit length in a strand of a given conductor (the ith turn) is [8]
√
I02 ρ 2 2πρ 2 d0 I
P0 = √ ψ1 (ζ) − H ψ 2 (ζ) , ζ = , I0 = . (16)
2πδd0 δ δ N0
116 Luo et al.

It is assumed that the external magnetic field across every turn is constant, thus proximity losses
can be calculated as the sum of contributions due to the internal magnetic field (H int ) and to the
external one (H ext ) [7]. Losses of a strand at the radius r of the conductor can be expressed as
√
I02 ρ 2 2πρ ¡ 2 2
¢
P0 = √ ψ1 (ζ) − Hint (r) + Hexti ψ2 (ζ) (17)
2πδd0 δ
where
√ √ √ √
ber(ζ/ 2)bei0 (ζ/ 2) − bei(ζ/ 2)ber0 (ζ/ 2)
ψ1 (ζ) = √ √
[ber0 (ζ/ 2)]2 +[bei0 (ζ/ 2)]2
√ √ √ √ (18)
ber2 (ζ/ 2)ber0 (ζ/ 2) + bei2 (ζ/ 2)bei0 (ζ/ 2)
ψ2 (ζ) = √ √
[ber(ζ/ 2)]2 +[bei(ζ/ 2)]2
ber() and bei() are the zeroth-order Kelvin functions, and ber2 () and bei2 () are the second-order
Kelvin functions. ber0 () and bei0 () are the first derivatives of ber() and bei(), respectively. These
functions can be calculated by using Bessel function:
h ³ 3πi ´i
ber (x) = Re J0 xe 4
h ³ 3πi ´i
bei (x) = Im J0 xe 4
h ³ 3πi ´i
ber2 (x) = Re J2 xe 4
h ³ 3πi ´i
bei2 (x) = Im J2 xe 4 (19)
· µ ¶ µ µ ¶¶¸
0 dber (x) √ 1 i √ 1 i
ber (x) = = Re 2 − J 2x − +
dx 2 2 1 2 2
· µ ¶¸
0 dber (x) √ 1 i √ 1 i
bei (x) = = Im 2( − )J 2x(− + )
dx 2 2 1 2 2
where Jv is the vth-order Bessel function of the first kind.
The density of eddy current losses at the radius r of the ith turn conductor can be obtained by
multiplying the eddy current losses density of a given strand at radius r, which is Pπr
0 (r)
2 by packing factor
0
kπr 20 kr 20
β= =
πrc2
.
Here packing factor β indicates the ratio of the total cross-sectional area of all strands
rc2
in the conductor to the cross-sectional area of the whole conductor.
√
dPi P0 (r) I02 ρβ 2 2ρβ ¡ 2 2
¢
= 2 β= √ ψ1 (ζ) − Hint (r) + Hexti ψ2 (ζ) (20)
dS πr0 2 2π 2 δr0 δr0
From (9), (15) and (20), we can obtain power loss per unit length in the given conductor by
integrating the density of eddy current losses
Z 2π Z rc µ ¶ √
dPi I 2ρ I 2 ρβ tan2 θ 4 2πρβI 2 rc2 Hexti
2
Pi = rdrdϕ = √ ψ1 (ζ)− √ 1+ ψ2 (ζ)− ψ2 (ζ) (21)
0 0 dS 2πδN0 d0 2πδd0 3 δd0 I2
Therefore, the power dissipation of the whole coil is the sum of power dissipation of all turns
NT µ ¶ √ NT
X I 2 ρNT lT I 2 ρβNT lT tan2 θ 4 2πρβI 2 lT r2c X 2
Hexti
P= Pi lT = √ ψ1 (ζ)− √ 1+ ψ2 (ζ) − ψ2 (ζ) (22)
2πδN0 d0 2πδd0 3 δd0 I2
i=1 i=1
The AC resistance of the coil can be expressed by
√ √ µ ¶ √ NT
2P 2ρNT lT 2ρβNT lT tan2 θ 8 2πρβrc2 lT X 2
Hexti
Rac = 2 = ψ1 (ζ) − 1+ ψ2 (ζ) − ψ2 (ζ) (23)
I πδN0 d0 πδd0 3 δd0 I2
i=1
Progress In Electromagnetics Research C, Vol. 51, 2014 117

4. EXPERIMENTAL RESULTS

Since it is impossible to measure AC resistance of a winding directly due to the influence introduced
by stray-capacitance, a lumped-parameter equivalent circuit model of inductors is used [10], as shown
in Fig. 6(a). L is the nominal inductance, Rac the AC resistance, and C the stray-capacitance. Most
LCR meters measure an equivalent series reactance Xs and an equivalent series resistance (ESR) Rs as
shown in Fig. 6(b):
Rac 1 − ω 2 LC − CRac
2 /L
Z= + jωL =Rs + jXs = |Z| ejθ (24)
(1 − ω 2 LC)2 + (ωCRac ) 2
(1 − ω 2 LC)2 + (ωCRac ) 2

The quality factor of the coil is given by

¯ ¯
|Xs | 2 /L)¯
¯ωL(1 − ω 2 LC − CRac
Q= = (25)
Rs Rac
The stray-capacitance can be calculated by
1
C= 2 2 (26)
4π fself L
Tests were performed using a HIOKI 3532-50 LCR HITESTER equipped with a HIOKI 9262 test
fixture. Two coils that were tested have a same radius of 380 mm. The first coil has 56 turns while the
second has 20. The air gap in the middle of both Helmholtz coil is 327 mm. The diameter of litz-wire
is 2 mm, and there are 180 strands in one bundle. Each strand has a diameter of 0.1 mm and the pitch
of the litz-wire is 16 mm.
R ac L Rs Xs

(a) (b)

Figure 6. (a) Lumped-parameter equivalent circuit. (b) Equivalent series circuit.

The inductances and self-resonant frequencies of both coils is measured: L1 = 2.751 mH, L2 =
0.448 mH, fself 1 = 830 kHz and fself 2 = 1907 kHz. The self-resonant frequencies were measured at zero
phase of the impedance.
Thus stray-capacitance can be calculated using (26): C1 = 13.36 pF and C2 = 15.55 pF.
When the frequency is above self-resonant frequency, the impedance is capacitive. In the WPT
system, the transmitting coil and the receiving coil form an inductive link, so the frequency of the WPT
system must be below the self-resonant frequency.
Figure 7 shows the measured and computed ESR Rs as a function of frequency f . AC
resistances (23), self-resonant frequencies, inductances and stray-capacitance of the coils are used to
calculate Rs . For both coils, when frequency is higher than self-resonant frequencies, error increases
dramatically as the frequency increases and measured data becomes very unstable. This is due to the
limitation of the model in Fig. 6(a), which is based on the first self-resonant frequency of the coil. So
electrical properties of both coils are not in agreement with this model when frequency is much higher
than first self-resonant frequency.
Figure 8 shows the measured and computed Q as a function of frequency f . Measured Q is obtained
by dividing inductive reactance by Rs : Q = ωL Rs , where Ls is measured equivalent series inductance
s

(ESL).
At higher frequencies, the measured Q values are lower than calculated ones, especially in the
NT = 56 case. This is caused by the deviations between calculated inductance values and measured
ones, which increase as the frequency becomes higher. Since the calculated inductance values are bigger
than measured ones, the calculated Q values become bigger than measured ones. However, for the coil
118 Luo et al.

Figure 7. Measured and computed ESR Rs of both coils (NT = 56 and NT = 20) as a function of
frequency f .

Figure 8. Measured and computed quality factor of both coils (NT = 56 and NT = 20) as a function
of frequency f .

with 20 turns, smaller deviations in calculated and measured inductance values result in a less obvious
phenomenon. On the other hand, both inductance and resistance values of the 20 turns coil are smaller
than those of the coil with 56 turns, while the accuracy of the LCR meter is constant in this particular
measurement range. So for lower turns, errors in measured ESL data are larger, thus will cause greater
discrepancy in the frequency range around peak Q value.

5. CONCLUSION

In this paper, an analytical expression of the AC resistance of a single-layer air-core Helmholtz coil
wound by using litz-wire has been derived. This expression is based on the analytical expression for the
AC resistance of solid round wire winding and has taken the magnetic field distribution across winding
space into consideration. The accuracy of expression has been verified by comparing the calculated
results with experimental results. The theoretical results were in agreement with measured ones when
frequency is below the self-resonant frequency of the coil.
Progress In Electromagnetics Research C, Vol. 51, 2014 119

REFERENCES

1. Van Gossum, A. and M. Ibrahim, “Video capsule endoscopy: What is the future?,” Gastroenterol.
Clin. North Am., Vol. 39, 807, 2010.
2. Shiba, K., A. Morimasa, and H. Hirano, “Design and development of low-loss transformer for
powering small implantable medical devices,” IEEE Trans. Biomed. Circuits Syst., Vol. 4, 77–85,
2010.
3. Ma, G. Y. and G. Z. Yan, “Wireless powered microrobot for gastrointestinal detection,” ICMA’07:
Int. Conf. on Mechatronics and Automation, 1085–1089, 2007.
4. Chen, C. J., T. H. Chu, C. L. Lin, et al., “A study of loosely coupled coils for wireless power
transfer,” IEEE Transactions on Circuits and Systems II: Express Briefs, Vol. 57, No. 7, 536–540,
2010.
5. Chen, L., S. Liu, Y. C. Zhou, et al., “An optimizable circuit structure for high-efficiency wireless
power transfer,” IEEE Transactions on Industrial Electronics, Vol. 60, No. 1, 339–349, 2013.
6. Bartoli, M., N. Noferi, A. Reatti, and M. K. Kazimierczuk, “Modeling litz-wire winding losses
in high-frequencies power inductors,” Proc. IEEE Power Electronics Specialists Conf., 1960–1966,
Baveno, Italy, Jun. 24–27, 1996.
7. Ferreira, J. A., “Analytical computation of AC resistance of round and rectangular litz-wire
windings,” IEE Proc. B, Electr. Power Appl., Vol. 139, No. 1, 21–25, 1992.
8. Tourkhani, F. and P. Viarouge, “Accurate analytical model of winding losses in round litz wire
winding,” IEEE Trans. Magn., Vol. 37, No. 1, 538–543, 2001.
9. Nan, X. and C. R. Sullivan, “An equivalent complex permeability model for litz-wire windings,”
IEEE Trans. Ind. Appl., Vol. 45, No. 2, 854–860, 2009.
10. Wojda, R. P. and M. K. Kazimierczuk, “Winding resistance of litz-wire and multi-strand inductors,”
IET Power Electronics, 2012, Vol. 5, No. 2, 257–268, 2012.
11. Wojda, R. P. and M. K. Kazimierczuk, “Analytical optimization of solid–round-wire windings,”
IEEE Transactions on Industrial Electronics, Vol. 60, No. 3, 1033–1041, 2013.