Fundamental Limits and Design Guidelines For Miniaturizing Ultra-Wideband Antennas

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Fundamental Limits and Design Guidelines

for Miniaturizing Ultra-Wideband Antennas


B. A. Kramer, C.-C. Chen, M. Lee, and J. L. Vo/akis
ElectroScience Laboratory, Dept. of Electrical and Computer Engineering, The Ohio State University
1320 Kinnear Rd., Columbus, OH 43212, USA
E-mail: {kramer.115.chen.118. lee.2070, volakis.1}@osu.edu

Abstract
Fundamental physical limitations restrict an antenna's performance based on its electrical size. These fundamental limits are
of the utmost importance, since the minimum size needed to achieve a particular figure of merit can be determined from
them. In this paper, the physical limitations of antennas are reviewed in general, with particular emphasis on impedance
matching as it relates to ultra-wideband (UWB) antennas (high-pass response). Additionally, the use of antenna
miniaturization techniques to approach these limits is discussed. Using a spiral antenna as an example, guidelines are
presented for miniaturizing UWB antennas.

Keywords: Electrically small antennas; antenna theory; impedance matching; loaded antennas; log spiral antennas; antenna
miniaturization; ultra-wideband antennas

1. Introduction of the antenna becomes prohibitive for most platforms. Therefore,


the antenna size must be reduced, making it electrically small over
the lower portion of its operational bandwidth. As the electrical
I t is not unusual to employ several antennas to cover a large
operational bandwidth that encompasses many different com-
munication systems. The advantage of this approach is that each
size of the antenna decreases, its performance will also begin to
degrade. For an antenna engineer, it is important to know how
small the antenna can be in order to achieve a desired performance
antenna can be designed to meet the specific requirements of a
specification at fi. This requires an understanding of the
communication system. However, integrating numerous antennas
into a platform (such as an aircraft) is challenging, because of fundamental limitations on directivity, radiation Q, and impedance
payload restrictions, limited space, cost, and electromagnetic com- matching for antennas.
patibility/interference (EMC/EMI) issues. Therefore, there is sig-
nificant interest in ultra-wideband (UWB) antennas that are physi- In this paper, we examine and review the fundamental physi-
cally small (in aperture and profile), and that have sufficient band- cal limitations on antennas, with particular emphasis on impedance
width to cover the desired operational bandwidth. In most cases, matching as it pertains to UWB antennas. We begin by discussing
the desired bandwidth can be anywhere from 10:1 to more than the limitations on directivity and radiation Q. Using the concept of
100:1. the minimum-Q antenna, limitations on impedance matching are
then discussed for narrowband (bandpass response) and UWB
For the antenna engineer, an antenna that belongs to the class antennas (high-pass response). In the remainder of this paper, we
of frequency-independent antennas [1, 2] is an obvious choice for discuss how to approach these limits by means of antenna minia-
achieving a large operational bandwidth. For this type of antenna, turization. Specifically, we discuss the concept of antenna minia-
the highest frequency of operation, fh' is in theory not limited. turization, and how it can be used to enhance the performance of
That is, fh can always be increased by adding more turns to the electrically small antennas.
center of a log-spiral antenna. However, in practice, fh is usually
We remark that this topic has been previously addressed in
limited by the size of the waveguide or transmission line feeding
[3]. Specifically, [3] discussed several practical techniques to
the antenna structure. On the other hand, the lowest frequency of
reduce size and improve impedance matching for specific anten-
operation, fi, is always limited by the maximum dimension of the
nas. In contrast, this paper focuses on the theory that defines the
aperture. Therefore, in the absence of any size restrictions, a fre- limits of miniaturization for wideband antennas. We also discuss
quency-independent antenna is theoretically capable of covering tradeoffs associated with broadband antenna miniaturization, and
any operational bandwidth. Of course, in practice, there are size provide basic guidelines for size reduction using a spiral antenna as
restrictions based on the size of the platform and the lowest oper- an illustrative example. However, we do not provide details of the
ating frequency, fi. In many cases, fi can be in the VHF band or implementation of the various miniaturization techniques (e.g.,
in the lower part of the UHF band, where the wavelength, A, is on dielectric loading or inductive loading). Instead, we approach the
the order of several meters. At such frequencies, the physical size miniaturization of broadband antennas from a general viewpoint.

IEEE Antennas and Propagation Magazine, Vol. 51, No.4, August 2009 ISSN 1045-9243/20091$25 ©2009 IEEE 57
As such, the presented guidelines should be applicable to any of With proper choice of spherical-wave mode coefficients, it
the chosen miniaturization techniques. was shown that the highest possible directivity that can be achieved
using spherical-wave modes of order n ~ N is [8, 9]

N
2+2N.
2. Fundamental Physical Limitations Dmax = L(2n+l)=N (2)
n=l

The fundamental limits on antennas are determined from the Therefore, regardless of antenna size, the directivity is unbounded
physical characteristics of spherical modes ("modes of free space") as long as the number of modes used to express the field is unre-
that can be used to express the fields produced by an antenna. This stricted. The reason for this unbounded directivity stems from the
method involves enclosing the entire antenna structure, its trans- inherent assumption that any or all of the spherical-wave modes
mission line and generator, inside a sphere of radius a. The field can be excited, regardless of antenna size. Stratton [5] was credited
external to the sphere is expressed using a complete set of orthogo- by Chu as being the first to demonstrate the impracticality of
nal TE and TM spherical modes, which propagate radially outward. achieving an arbitrarily large directivity. In his unpublished notes,
Depending upon which modes are excited (modal coefficients), Stratton demonstrated that the series representing the source distri-
and the electrical size, ka ( k = 21C/ A) of the sphere, limits on bution diverges as the directivity of the system increases indefi-
directivity, radiation Q, and impedance matching can be calculated nitely. Therefore, it is physically impossible to excite all of the
under ideal conditions 1• These limitations are typically discussed modes with a finite-size antenna. However, there still wasn't a
with regards to electrically small antennas'. However, in their most clear physical reason for limiting the number of spherical modes
general form, these limitations apply to an antenna of arbitrary size. that could be excited. It was Chu [5] who introduced the concept of
Below, we review the fundamental limits on antennas, and discuss the quality factor, Q, to define a cutoff for each spherical mode. In
how they apply to UWB antennas. doing so, he provided a physical basis for limiting the number of
modes used in determining the maximum directivity of a given
antenna size. Before we review how Chu used the concept of Q to
limit the directivity, we first discuss the fundamental limit on an
2.1 Directivity antenna's Q.

In the early 1900s, a discussion began about whether it was


possible for an aperture to have infinite directivity if the amplitude
and phase of the current distribution was properly chosen. In one of 2.2 Quality Factor, Q
the first published articles on this subject, Schelkunoff [7] derived
a current distribution for a linear array that could achieve super-
The quality factor, Q, is a fundamental concept in physics for
directivity. That is, he determined the current distribution which
characterizing an oscillating system. In general, the Q of an oscil-
resulted in a higher directivity than if the array was excited with
lating system is defined to be the total energy content of the oscil-
uniform amplitude and linear phase (constant or progressive across
lator at a given time, divided by the energy loss per radian [12]:
the aperture). In later works, it was shown that there is no mathe-
matical limit to the directivity of an antenna for a given size [5, 8,
9]. This was demonstrated by first deriving an expression for the (3)
directivity in terms of the spherical-wave mode coefficients Amn
(TM mn modes) and Bmn (TE mn modes) [10]. Specifically, the
That is, Q tells us how the energy of the oscillating system dies out.
directivity can be expressed as From Equation (3), and assuming an oscillating system having ini-
tial energy Wo at time t = 0, the total energy content of the system
f ±(AmnQ~IAnmI2 +AnmQnlBnmn at any time t is given by W = Woe-mtlQ [12]. Therefore, the smaller
the Q, the more quickly the energy of the oscillation dissipates.
D = larger of n=lm=O ,(1)

f ±(AnmQn IAnm + AnmC!n jBnm


n=lm=O
2
1
2
1 )
This means that an antenna with a low Q can quickly radiate
energy, making it a very effective radiator. Therefore, the Q is of
significant physical importance for quantifying antenna perform-
ance.
where A"m' Q~, and Qn are functions defined in [11]. The con-
stant A"m is independent of ka, whereas Q~ and Qn are compli- The first prominent publication on antenna Q and its use in
cated functions of ka involving spherical Bessel, Neumann, and describing the fundamental limitations of antennas was by H. A.
Hankel functions. Note that in Equation (1), it is assumed that the Wheeler, in 1947 [4]. In [4], Wheeler's analysis was focused on
coordinate system can always be oriented such that the maximum extremely small antennas (ka « 1), which he approximated either
field strength occurs at () = 1C/2, ¢ = 0° [8, 10]. as a lumped capacitor, or as a lumped inductor connected to a
resistor representing radiation resistance. Using this simple theory,
Wheeler was able to show that the ratio between the antenna's
resistance and reactance (energy radiated!dissipated and energy
stored) - which he referred to as the power factor - had an upper
1The conditions are that no energy is stored or dissipated inside the bound, determined only by the antenna's electrical size, ka. A year
sphere; Further, the input impedance is purely resistive at a given later, a more comprehensive analysis was carried out by Chu [5],
ka [5]. which was applicable to an antenna of arbitrary size. In Chu's
2The term "electrically small" is used to refer to any antenna with analysis, he determined a lower bound on the antenna's Q using the
an electrical size, ka, that is less than one [4]. radiation Q associated with each TE and TM spherical mode. That

58 IEEEAntennasand Propagation Magazine, Vol. 51, No.4, August 2009


&r
c~&r calculation of the Q for each spherical mode was based on an
n 2n-3 approximation that helped facilitate the tedious computation. The
• ••.•.••••.•••• ~ R=1J approximation made by Chu and repeated by Harrington was to
approximate the equivalent circuit for each mode using a second-
ZTM-----+ order series RLC circuit that had the same frequency behavior in

__-------...-.--- f
mn
the neighborhood of the operating frequency [5]. Later, Collin and
Rothschild [14] published an exact theory for calculating the
radiation Q using a field-based technique, rather than equivalent
(a) circuits. Subsequently, Fante [11] generalized Collin's technique to
include both TE and TM modes. The exact equation for the mini-
&r
mum Q as given by Fante (when an arbitrary number of modes are
c~~ present) is [11]
2n-l 2n-5
00 00

L~pr
L:[
n=l
a;Qn+b;Q~ J ._..n=L:[
__l
a;Q~ +b;Qn J
_
n Q ~larger of , (4)
f[a; +b;]
n=l
f[a; +b;J
n=l
(b)
a; == m=O
n n
Figure 1. Equivalent circuits for the (a) TM mn and (b) TE mn where L: A"m /Anm b; == L: A"m IBnm/
I
2
,
2
, and A"m is a con-
modes of free space [5, 8]. m=O
stant (see [11] for definitions).

When only the lowest-order modes are present (TM oI and


is, Chu calculated the Q of each mode, assuming that no energy
TE oI ), Equation (4) reduces to [15]
was stored inside the radian sphere by the source distribution that
excited it. Therefore, for any antenna that fits inside the sphere, the
minimum possible Q that can be achieved is limited by the Q of the
(5)
spherical modes it excites, because any energy stored inside the
sphere can only increase the Q.

In order to determine the Q of each spherical mode, Chu had where M = 1 refers to the excitation of either the TE oI or TM ol
to separate the radiated energy (propagating) from the energy asso- mode (linearly polarized field), and M = 2 stands for equal excita-
ciated with the near field (non-propagating) external to the sphere. tion of both the TE oI and TM oI modes (the field can be either
Since the total system energy is infinite and the near-field compo- circularly or linearly polarized). It is remarked that for M = 2 ,
nents are nonlinear [5], this is not a trivial matter. Chu did this by Equation (5) provides the absolute lowest Q to be achieved by an
transforming the field problem into an equivalent circuit problem, antenna of a given size.
by replacing the space outside the sphere with a number of inde-
pendent equivalent circuits. Each equivalent circuit represented an
orthogonal spherical-wave mode used to describe the external
field. The equivalent circuits were obtained from the wave imped- 2.3 Maximum Directivity for a Specified
ance of each mode, which completely described the propagating
and non-propagating energy associated with each mode [5]. Using
Radiation Q
the recurrence formulas for spherical Bessel functions, Chu
obtained a partial-fraction expansion of the wave impedance for Previously, it was discussed that there was no mathematical
each mode. The resulting expansion was then interpreted as a high- limit for antenna directivity, as long as the number of excited
pass RLC ladder network, as shown in Figure 1. The network con- spherical modes was unbounded. To provide a means for restrict-
sisted of a cascade of series capacitances and shunt inductances, ing the number of modes, Chu was the first to use the concept of Q
with the resistive element at the end of the network representing to quantify the practicality of exciting a given spherical mode. The
the radiated (propagating) energy. The Q could then be calculated first attempt to define a practical limit for the directivity relied on
using the equivalent circuit for each mode by summing the energy the observation that the Q of spherical modes of order n is consid-
stored in each of the inductors and capacitors. In doing so, Chu ered "high" for lea < n. It can therefore be assumed that modes of
demonstrated that the lowest radiation Q can only be obtained if order n > ka are not easily excited, and do not significantly
the lowest-order spherical modes (TE ol and TM ol ) were excited contribute to the antenna's fields [5, 8]. Under this assumption, the
[5]. However, there were two issues with Chu's work that were cutoff point is taken to be ka = n . Further, recalling that the maxi-
later addressed by various authors. mum directivity that can be achieved using spherical modes of
order n ~ N is given by Equation (2),a limit for antenna directivity
First of all, Chu's analysis only considered the presence of can be obtained by substituting N = lea into Equation (2):
either TM or TE modes. Therefore, it was not complete. With this
in mind and following Chu's method, Harrington considered the Dnormal = (lea )2 + Zka . (6)
case where both TE and TM modes were present. In doing so,
Harrington showed that the absolute lowest achievable radiation Q This limit is commonly referred to as the "normal directivity"
was obtained by concurrent and equal excitation of the TM o1 and limit. We observe that as lea becomes large, the normal directivity
TE ol modes [13]. The second issue with Chu's work was that his reduces to that obtained by a uniformly illuminated circular aper-

IEEE Antennas and PropagationMagazine, Vol. 51, No.4, August 2009 59


ture of radius a [8, 9]. It is also remarked that Equation (6) is con- and a resistive generator-was initiated by Bode [17] . Bode consid-
sidered to be the maximum directivity achievable without incurring ered a two-element RC or RL load, and determined the maximum
a "high" Q [10]. possible bandwidth for a given maximum tolerable reflection coef-
ficient , I'0' within the pass-band, as illustrated in Figure 2. It is
ill deriving the normal directivity and minimum Q, Chu [5] important to note that Bode only considered a matching network
noted that both could not be achieved simultaneously. That is, the with an infinite number of stages. Later, Fano generalized Bode's
spherical-mode coefficients required to maximize the directivity work by extending it to include arbitrary loads and an arbitrarily
where not the same as those that minimized Q. This implies a fun- complex matching network [18, 19]. In both Bode's and Fano's
damental limit on the product or ratio of the directivity and Q. work , the system response was arbitrary. However, in the literature
Therefore, Chu proposed to maximize the ratio D/ Q by simultane- (especially in antenna theory), their work was presented for the
ously maximizing the directivity while minimizing Q. This ratio is bandpass response [20-22], which is suitable for narrowband
of important significance, because it provides the optimal tradeoff antennas. Here , we discuss impedance-matching limitations for
between directivity and Q for an antenna of given size. Addition- both the narrowband (bandpass) and broadband (high-pass) cases.
ally, it can be used to determine the maximum achievable directiv- We begin by discussing Fano 's method for determining imped-
ity for a given Q, or the minimum achievable Q for a given direc- ance-matching limitations for an arbitrary load. Due to the com-
tivity. ill maximizing this ratio, Chu only considered the TM plexity of Fano's method, we will restrict our discussion to a high
modes, making his analysis only applicable to the omnidirectional level overview of his approach.
linearly polarized antenna' , Fante [10] extended his work to
include both TE and TM modes, and calculated the maximum Consider the problem of designing an optimum lossless
D/Q ratio for the directional antenna case . Later, Geyi [16] revis- passive network to match an arbitrary load impedance to a resistive
ited Fante's work to correct an error in Fante's analysis. Geyi then generator, as illustrated in Figure 3. Fano solved this problem in
recalculated the maximum D/Q ratio for the omnidirectional and general for any load impedance that could be represented using a
finite number of passive circuit elements. The basic idea of Fano 's
directional antenna cases by means of an optimization process
approach was to tolerate a certain amount of mismatch between the
subject to constraints found in [16]. Basically, the optimization
load and generator, such that the bandwidth (or matching area [23])
process required adjustment of the amplitude and phase of the
was max imized. The first step in Fane's method was to simplify
spherical expansion coefficients in Equations (1) and (4) until the
the problem by replacing the load impedance with its Darlington
ratio was maximized. For the directional antenna, Geyi obtained
the following expression:

max (D)I =2f 2n+l . (7)


\rl
Q dir n=l Qn +Q~
1
For the omnidirectional antenna, the maximum D/Q ratio was
found to be [16]

(8)

where P~ (0) is the Legendre polynomial of the first kind. The sig-
IrQI-~"""""""""""1!-------!
nificance of the above expressions is that they demonstrate the
conditional nature of defining a limit for the directivity. For i
instance, if an antenna engineer wants to design an antenna that I I
meets the minimum Q limit of Equation (5), then the max imum
directivity, D max , must be 1.5 for M =1, and 3 for M =2 [5, 9].
If a directivity of 1.5 or 3 is not sufficient, then according to Equa- Figure 2. A possible bandpass response for the reflection coef-
tions (7) or (8), the antenna Q must be greater than the lower bound ficient that illustrates the Bode-Fano criterion.
given by Equation (5). It is also important to emphasize that this
result is applicable to an antenna of arbitrary size .

2.4 Narrowband and Broadband Lossless


Impedance Matching Passive
Matching
The classic problem of designing a passive reactive network Network
to maximize the impedance bandwidth between an arbitrary load

3The term omnidirectional refers to an antenna that radiates into


r
both hemispheres, whereas the term directional refers to an antenna Figure 3. A matching network for an arbitrary load impedance
that radiates only into one hemisphere. [19J.

60 IEEEAntennasand Propagation Magazine, Vol. 51, No.4, August 2009


Network N The resulting expression relates the return-loss function (left-hand
-- - - - -- --- - - - - -- - - - - - - --. side) to the zeros and poles of the load and matching network
I

(right-hand side). Specifically, the Agk+1 coefficients come from


Network N" Network N' ......----.
LossIes. ReactIYe
PassIve
Matching
Networt<
Determined
the Taylor-series expansion of the return-loss function, In(!fIr!) ,
Network byz..
and depend only on the load [18]. They are given by
~
IL I
~ ~

AO
2k+1
=_1_ (," S -{2k+l) -
2k + I L.:- 01
'" S-(2k+l)
L.- pI
J '
(12)
1 I

Figure 4. A matching network and the Darlington equivalent of


where soi and S pi are the zeros and poles of the load impedance,
the load impedance in cascade (19).
respectively. The last term on the right-hand side of Equation (11)
is the unknown zeros, sri' of the matching network that lie in the
equivalent circuit, as shown in Figure 4. He then normalized the right half plane.
impedance, making the resistances equal to unity. The resulting
network could then be described by two reactive networks con- The optimum tolerance of match is found by determining the
nected in cascade to form a single two-port network N, terminated
at each port with a 1n resistor. Since the reflection coefficient, I" ,
function Irdthat simultaneously maximizes the integral relations.
of a lumped-element reactive network must be a ratio of two real In other words, Ird must maximize the "matching area" given by
polynomials in the complex frequency variable S = 0'+ jOJ [18], F the integral relations. For a given load, this is accomplished by
must have the following form: proper selection of the zeros, sri' of the matching network. Ideally,
the matching area is maximized when Ird has a rectangular-
(9) shaped response (see Figure 2 or Figure 5). For the rectangular-
shaped response, the left-hand side of Equation (11) can be evalu-
ated analytically, making a closed-form solution possible. Of
where K is a real number, the sOm are the zeros, and the S pm are course, the ideal rectangular-shaped response can only be obtained
if the matching network has an infinite number of stages. When the
the poles. Knowing the form of I", Fano proceeded to determine
the zeros and poles that would maximize the bandwidth subject to matching network has finite complexity, the optimum is found Irll
the tolerance of match, r o- However, the selection ofthe zeros and by approximating the ideal rectangular response with a set of basis
poles had to satisfy various constraints (see [18] for details). For functions. In [18], Fano used Tchebysheffpolynomials to represent
instance. r had to satisfy constraints based on the behavior of the Ird and to obtain a solution for an RLC load with a finite number
load impedance. Also, the network, N, and therefore F had to be of matching stages. We now briefly review this solution, which can
physically realizable. Essentially, these constraints placed limits on be used to determine the maximum possible bandwidth for a nar-
how well the load impedance could be matched. To satisfy these rowband antenna.
constraints, Fano developed a series of integral equations involving
the logarithm of the magnitude of the reflection coefficient. The
form of these equations depends upon the type of response (i.e.,
bandpass, high-pass, low-pass, etc.) for the overall system, and the
response of the load.

For an overall high-pass system response, the equations given


by Fano involved an integral ofthe form

(10)
1
for k = 0,1,..., N -1, with N representing the multiplicity of the
transmission zeros associated with the load", Each integral is
evaluated over the frequency spectrum of a return-loss function,
In(!flrd), which is used instead of Ird
for purely mathematical
reasons. By using the calculus of residues, Fano evaluated Equa-
tion (10) to obtain the following expression for a high-pass
_ _ •• •• •••• •• • •••• •••••• • L - _
response:

l""OJ- 2(k+I) In ( _I )dOJ=(_I)k!!..[AO __2_,"s-{2k+I)] •


.b Ird 2 2k+1 2k+17 rr i
io
(11)

4The integral equations can be written terms of either r I or I'2 . Figure 5. The ideal high-pass reflection-coefficient response.

IEEE Antennasand Propagation Magazine, Vol. 51, No.4, August 2009 61


2.4.1 Fano-Bode Limit for the important to note that the center frequency is not restricted in any
Bandpass Response I
way. That is, regardless of antenna size, any [r 0 is achievable as
long as one is willing to reduce Bn . However, for the high-pass
To begin, the fractional bandwidth for the bandpass response response, we will show that the electrical size will determine the
is defined as Ir I.
lowest possible 0 That is, for the high-pass response, the funda-
mental tradeoff is between Ir0 I and the cutoff frequency, OJe , or
(13)
"cutoff size" (ka)e.

where OJp2 and OJpl are the high and low edge-band frequencies
for which Irl s Ir0I (see Figure 2). Additionally, the center fre-
quency of the bandpass response is defined to be ~OJp2OJpI.
2.4.2 Fano-Bode Limit for the
Fano's solution for the bandpass response was in terms of a set of
equations [18]:
High-Pass Response

For antennas having a continuous high-pass response, such as


13 _ 2sin(~J (14)
a spiral or some other frequency-independent antenna, the imped-
Q n - sinh(a)-sinh(b)' ance-matching limitations for the band-pass case are not relevant.
Here, Fano's work is adapted to the high-pass matching case,
which is more applicable to broadband antennas. Similar to the
tanh (na) tanh (nb) bandpass limitations, the impedance-matching limitations for the
(15)
cosh] a) = cosh(b) , high-pass case are obtained using the equivalent circuits of the
lowest-order spherical modes ( TEo I and TM oI) as the load
cosh(nb) impedance. Using Fano's method, a relationship between the cutoff
ro=-~~ (16) frequency, OJe , and the maximum tolerable reflection coefficient,
cosh(na) ,
Ir0 I, is obtained for three specific cases. Note that the same
where Q is the quality factor associated with the load, n is the approach was used in [26] to determine matching limits on UWB
number of matching stages", and the coefficients a and b are antennas. However, the cases considered in [26] were different
unknowns. Matthaei, Young, and Jones were among the first to than those considered here, and involved higher-order spherical
publish solutions to Fano's equations in the form of tables for the modes. All of the cases considered here involve only the lowest-
coefficients a and b [24]. However, their coefficients were depend- order modes.
Ir
ent upon the choice of 01. Lopez [20], using a methodology
For the first case, we consider the matching limitations
developed by Wheeler [25], provided a closed-form solution in
Ir
which his coefficients were independent of 01. Lopez's equation
obtained by connecting the equivalent circuit of the TE oI or TM oI
mode directly to the generator (no matching network, n = 1). In the
was as follows:
second case, we consider the matching limitations obtained when
both TE ol and TM ol modes are excited equally (lowest possible
(17) antenna Q). As in [27], it is assumed that the same current distri-
bution equally excites both modes, resulting in the series combina-
tion of their respective equivalent circuits. Using this series combi-
nation, the matching limitations are determined by connecting the
where the (an' bn ) coefficients were given in [20], and depend load directly to the generator. For the third case, we demonstrate
the potential improvement that can be obtained by including an
only on the number of matching stages. From Equation (17), it is
infinitely complex matching network (n = OC)) in the first case.
Ir I,
evident that for a given n and 0 B; is inversely proportional to
Q. Therefore, for a bandpass type antenna, the upper bound on B; We remark that for the first two cases (n = 1) it is not neces-
is obtained upon substitution of the minimum Q (Equation (5) with sary to use Fano's integral relations to determine the impedance-
M = 1). Essentially, this is analogous to replacing Fano's RLC matching limits, since the form of the reflection coefficient is
load with the equivalent circuit for the TM oI or TE ol (see Fig- known and there is no matching network. For these cases, we only
ure 1). need to select the generator resistance, R s ' to calculate the reflec-
tion coefficient looking into the load. For the first case, R, is set
Before we proceed to discuss broadband (high-pass) imped- equal to the load resistance of the equivalent circuit, which hap-
ance matching, it is useful to note an important difference between
pens to be the impedance of free-space, 1]0 (see Figure 1). For the
the bandpass and high-pass cases. For a given matching network
complexity, n, and load impedance, there is a fundamentallimita- second case, the series combination of the equivalent circuits
Ir
tion or tradeoff between B; and 01 for the bandpass case. It is
results in a load resistance equal to 21]0 [27]. Therefore, R, is set
equal to 21]0 in the second case. For the first two cases, the funda-
mental relation between OJe and Ir0I can now be determined by
5The n = 1 case corresponds to the antenna or load being connected calculating the reflection coefficient with respect to R, looking
directly to the generator (no matching network). into the load. The results are shown in Figure 6.

62 IEEE Antennasand Propagation Magazine, Vol. 51, No.4, August 2009


1
- r- , '-
..... , :: has two poles, s = -c ± j c , and no zeros. Using Equation (12), we
2a
"
.............. .......... ; ' ., have AP = 2a/c and Af =-( Ap)3/6. As Fano noted, to
0.8
" : ,, :
\.:
,
-, simultaneously maximize the matching area defined by each inte-
0.6 ..: "
: ,,~ :"
;
gral, the zeros of the matching network must be selected so that
"LJ s -n3 is as large as possible, while "s - .1 is kept as small as
~o '""'';'. ' . : \.
~ n

. ',. ... . '-.:. . ..


possible. This is accomplished by using a single real zero, s; = CFr
0.4
, "- [19]. Solving Equation (18) for CFr and substituting the result into
" :',
- - TM01 or TE o1' n =1 ' ';'"
:
.~
,
".
' .
~ .- ....~
Equation (19) eliminates CFr • The resulting equation is the cubic
0.2 polynomial
--- TM 01 orTEo1' n = CD

_ ._ ..TM
01
and TE
o1
'n =1
0 (22)
0 0.2 0.4 0.6 0.8 1
ka
Figure 6. The in-band reflection coefficient I'0 as a function of
with
the cutoff size (ka = OJea ) or cutoff frequency. The cutoff size is
c
the smallest antenna size for which 0 Irl sIr I.

To demonstrate the impact of an infinite-stage matching net- The cubic equation was solved using MATLAB to find the
work (the third case), we must solve the set of equations given by roots of OJe for a given K . However, for any given K, only one of
Equation (11) using the equivalent circuit of the TM ol or TE ol the roots of OJe is real. The solution is shown graphically in Fig-
mode as the load. For this load there are two transmission zeros at
ure 6 (for the case of n =co). We remark that in Figure 6, OJe was
zero, implying a multiplicity of two (N = 2). For N = 2, the inte-
gral relations obtained from Equation (11) are as folJows: multiplied by a/c to convert it to the more-useful parameter ka .
The curve in Figure 6 for the n = co case should be interpreted as
Ir
follows. For a given 01 , there is a corresponding kea defining
(18)
the smallest electrical size for which the pass-band reflection coef-
Ir
ficient can equal 01 . To achieve this cutoff size, the frequency
and response of I' must be rectangular. That is, below k.«, the magni-
tude of the reflection coefficient must be one, and, for all frequen-

.broo OJ -4 In (1)
ifJ dOJ = 2"
-Jr [
A3 2" -3]
0-'37 Sr i .
(19)
Ir
cies above kea, it is exactly equal to 0\' Therefore, the curve in
Figure 6 corresponding to n = co cannot be realized in its entirety.

Since the matching network has infinite complexity, the form of The results in Figure 6 can be used in conjunction with the
Irll is that of the ideal rectangular high-pass response (see Fig-
directivity limits to define the realized gain limit. For the cases
where only the first one of the lowest-order spherical modes was
ure 5). Therefore, the left-hand side of Equations (18) and (19) can excited, the directivity is limited to 1.5 ('" 1.76 dB) for all ka [5].
be evaluated analytically. For the right-hand side, we must deter- For the case where both TM OJ and TEO! modes were excited
mine the A coefficients, which depend upon the load. To begin, the
equally, the maximum directivity that can be achieved for all ka is
equivalent circuit for the TE ol or TM ol mode is first normalized
3 [13]. Assuming the antenna is lossless, the realized gain is cal-
to the free-space impedance, 170 ' The impedance, in terms of the culated using the results from Figure 6 and the corresponding
complex frequency s = jOJ, seen looking into the equivalent circuit directivity. The resulting realized gain curves are shown in Fig-
is then given by ure 7.

S2a2 +sac+c 2 In closing, some final comments about the impedance-match-


1
ZTM 01 (s) =-Z--:(s-:-) s2a2 + sac
(20) ing limits are in order. First, it is important to emphasize that the
TEO) limits are for a minimum -Q antenna . Therefore, for an antenna to
reach these limits, the antenna Q must approach the lower bound
The reflection coefficient, normalized with respect to 17, is then on Q. Second, the impedance-matching limits do not apply if active
given by circuit elements are employed, or if there is loss in the antenna's
structure (i.e., resistive loading or an absorber). That is, Fano's
method is only applicable to lossless passive matching networks.
(21) Furthermore, the inclusion of loss within the antenna's structure
permits the antenna Q to be lower than the minimum Q because
Chu only considered the lossless case in his analysis. In fact, one
where the plus sign refers to the TM ol mode, and the minus sign can achieve an arbitrarily low antenna Q using resistive loading, at
refers to the TEO! mode. Regardless of the mode, Equation (21) the expense of efficiency.

IEEE Antennasand Propagation Magazine, Vol. 51, No.4, August 2009 63


_.-,_.-.
. . .... , ....-'
5 .--- - ---r-- - - ,--- - ----,-- - - -,-- -::-:-:=-:-:;;I It is well known that the phase velocity, v p ' and the
characteristic impedance, Zo, seen by the guided wave are given
: -: ----:------
.

:. . /":""~ ~.-:
. ",;'
.
o by

-iii : /
,;//
~,.,...?

/:
:

vp
I 1
= fiC = ..r;;; '
~ -5 ..........:.../ ..../ ..;
'=
".
:I
I
I (23)
C) J: I
"Cl / : I
,~ -10 .. ........( ..;... j ... ...
'i; 1 : /'
! J : .
... I
i
:' - - TM01 or TE01' n =1 where L is the series inductance per unit length, C is the shunt
-15 ......./I .......:..... capacitance per unit length, and G is a geometrical factor". The
I
I !:
r.
-- - TM 01 orTE01' n = m phase velocity can therefore be controlled using the series induc-
'
I i :
.
_ ._ ..TM
01
and TE ' n
o1
=1 tance and shunt capacitance per unit length of the antenna. For the
biconical antenna, this means controlling the self-inductance of the
0.2 0.4 0.6 0.8 1 cone and the capacitance between the two cone halves. By doing
ka so, we can achieve the proper electrical delay to attain resonance ,
or to form a radiation band for a spiral antenna, regardless of
Figure 7. The lowest achievable cutoff frequency for a mini- physical size. As we will show, this will improve the antenna's Q.
mum-Q antenna in terms of the realized gain.
Techniques that can be used to mod ify the inductance and
capacitance of an antenna structure involve either material or reac-
3. Antenna Miniaturization tive loading. Material loading refers to the application of materials
that have f r > 1 and/or IJr > 1. This is the most generic approach ,
making it applicable to any antenna design . However, the material
Having already discussed the fundamental limits on radiation
density can make this approach prohibitive for applications that
Q, directivity, and impedance match ing for antennas, the remainder
require lightweight antenna designs . Additionally, material losses
of the paper focuses on how to approach the impedance-matching
can playa significant role in their applicability. For example, the
limit with regards to electrically small UWB antennas . Since the
frequency-dependent loss of currently available magnetic or mag-
impedance-matching limit is based on a minimum-Q antenna, it
neto-dielectric materials prevents their use above VHF. On the
can only be approached using a low-Q antenna design . The meth-
other hand, reactive loading refers to any method that enhances the
ods that can be used to approach this limit can be divided into two
self-inductance and/or shunt capacitance within the antenna's
categories: 1.) minimum-Q antenna designs; 2.) min iaturization of
structure. A classic example of inductive loading is the meandering
preexisting antenna designs to lower their Q. The first approach is
of the conductor forming the antenna, such as a meander-line
about designing an antenna that utilizes the entire volume of the
dipole. Not only is this approach lightweight, but it is also applica -
radian sphere such that there is no stored energy inside the sphere,
ble for any frequency range. However, for some antennas, it can be
and only the lowest spherical modes exist outside it (TM ol and difficult if not impossible to implement capacitive and/or inductive
TE ol ) [5]. Wheeler proposed two such antenna designs, the loading.
spherical inductor and the spherical-cap dipole, which could come
close to achieving this goal [6, 28]. A characteristic of these
designs is that they occupy a significant portion of the radian
sphere. Therefore, for applications that require conformal or low- 3.2 Dipole Antenna Example
profile antennas , the first approach is not attractive . The second
approach involves the miniaturization of a preexisting antenna To illustrate how miniaturization improves an antenna's Q,
design to improve its Q at frequencies where it is electrically small. we consider a wire dipole that fits inside a sphere of radius a. The
In the remainder of this paper, we will discuss this approach. We length of the dipole is thus equal to 2a . Under these constraints,
begin by discussing the concept of antenna miniaturization. Subse- the dipole will have its first resonance at lea '" 70/2 7, which can be
quently, we demonstrate how the miniaturization of a physically
reduced by increasing the self inductance of the wire. Without
small dipole can improve its Q. Using such concepts, we then
physically modifying the dipole, this can be readily accomplished
develop basic guidelines for miniaturizing UWB antennas .
by using a Method of Moments code, such as NEC [31]. In NEC, a
distributed or lumped impedance (parallel or series circuit) can be
assigned to each wire segment that forms the dipole. In this case,
3.1 Concept each segment was assigned the same distributed inductance, L,
(uniform loading) thereb y increasing the inductance per unit length
The concept of miniaturization involves reducing the phase
velocity of the wave guided by the antenna structure to establish
resonance or coherent radiation when the antenna is electrically
61n general,the characteristic impedance of an antenna is frequency
small. To illustrate how this can be accompl ished, we use the anal-
ogy between an antenna and a transmission line. Consider a center- dependent, unless the geometry naturally scales with frequency (a
fed infinite biconical antenna - a type of a spherically radial constant geometrical factor) .
wavegu ide - guiding a spherical wave [7]. This is analogous to an 7The first resonance of a dipole occurs at lea = 70/2 only if the
infinitely long uniform transmission line guiding a plane wave [29, current distribution is perfectly sinusoidal and the wire radius is
30]. infinitesimally small.

64 IEEE Antennas and Propagation MagaZine, Vol. 51, No.4, August2009


12 r-- - - - - - - - ,- - - - -- --r-- - - - - - ,
4. Broadband Antenna Miniaturization
11

10 We now proceed to discuss broadband antenna miniaturiza-


........ .. .. .. .. .. .. ; .. tion in general. Our goal is to illustrate some simple guidelines that
9
antenna engineers can use to miniaturize a broadband antenna,
.............. .... ... : .."r
8 regardless of which miniaturization technique they choose to
7 employ. To demonstrate these guidelines, we use a wire log-spiral
u,
antenna (see Figure 12), and the inductive loading technique previ-
::i: 6 ..: : .
ously used for the dipole antenna. Examples of implementing
5 inductive loading can be found in [33], where lumped surface-
mount inductors were utilized, and in [34], where the spiral was
inductively loaded by coiling its arm such that it resembled a heli-
3 cal waveguide. In this section, our focus is on demonstrating the
2
concepts used in designing these antennas, rather than on a par-
ticular design. As a first step, we discuss how much the antenna
50 100 150
should be miniaturized to minimize the Q at a given ka . We then
Inductance per Un it Length (jJ.H/m) examine how the loading profile impacts antenna performance, and
we discuss the benefit of equal inductive and capacitive loading.
Figure 8. The miniaturization factor as a function of the wire
inductance per unit length for a dipole antenna.

4.1 Optimal Miniaturization Factor


For any type of antenna, it is important to know how much
the phase velocity must be reduced to minimize the antenna Q at a
given ka . For a bandpass or narrowband antenna, the phase veloc-
of the dipole. In doing so, the resonant frequency was shifted to a ity needs to be reduced such that the antenna is self-resonant at the
lower frequency as the inductance was increased. The ratio of the desired ka . However, for a broadband antenna, such as a spiral,
resonant frequency for the unloaded case (L = 0) to the loaded determining the optimal miniaturization factor, mopt ' is not as
case (L > 0) is defined herein as the miniaturization factor, m straightforward. Consider a spiral antenna (see Figure 12) that fits
f:
(m = es / fires). Figure 8 shows the miniaturization factor as a inside a sphere of radius a. From radiation band theory [35], we
know that the spiral radiates primarily from an annular band with a
function of the inductance per unit length. Note that we use this circumference that is one wavelength, A . For a circular spiral, the
approach to miniaturize the dipole because it does not alter the diameter of this radiation band is then D = A/lr , implying that the
physical structure of the antenna. Therefore, it is a very generic
smallest electrical size a spiral can have and still support a radia-
approach that can be used to demonstrate the impact of reducing
the phase velocity. tion band is ka = (2lr/A){D/2) =1. Now, consider a frequency,
~, for which a spiral antenna cannot establish a radiation band
To observe the impact of miniaturization on Q, we calculated within its aperture because it is physically too small (ka < 1). To
the dipole Q at several values of ka below resonance. To do so, we establish the radiation band within the spiral aperture, the phase
used the following expression [32]: velocity needs to be reduced at the very least by a factor of

fo radiation band before miniaturization

In Figure 10, the Q is plotted as a function of the miniaturization


factor for the various values of ka. For each ka , the dipole Q was
normalized using the Q of the unloaded dipole (the m = I case) to
make all curves viewable on the same plot. From Figure 10, it is
evident that the Q decreased as m increased until a minimum was
reached. The value of m for which the minimum Q occurred corre-
sponded to the miniaturization factor required to make the dipole
resonate at a given ka . This can be better depicted by re-plotting
the data using the effective electrical size, kma =mka , as shown in
Figure 9. Recalling that the unloaded (m = 1) dipole resonates
when ka is slightly less than lr/2, it was evident that the mini-
mum Q occurred when kma =1.5 . Of course, this verifies our ear- fo radiation band after uniform loading
lier statement that the Q is minimized when the dipole achieves Figure 12. An illustration of the effect of miniaturization on the
resonance. Therefore, we can improve the Q of a physically small location and size of a radiation band. The wire log-spiral is six
dipole by increasing the electrical size (reducing the phase veloc- inches in diameter, with an expansion ratio t = 0.525 (a
ity). growth rate of a '" 0.1), and an angle is = lr/2.

IEEE Antennas and Propagation Magazine, Vol. 51, No.4, Augusl2009 65


ml = 1/k}a. However, we could also establish the radiation band As a specific example of the above statement, let us uni-
closer to the center of the spiral by reducing the phase velocity by formly load the spiral in Figure 12 with an inductance per unit
any factor, m2' which is larger than mI' The question therefore is, length of 5 JlH (m = 2.21). The resulting Q is shown in Figure 13.
"Which miniaturization factor will minimize the Q at a given ka?" As seen, above 600 MHz, the miniaturized spiral Q was signifi-
To answer this question, we miniaturize a spiral antenna (see Fig- cantly higher because the radiation bands were electrically smaller.
ure 12) using the same inductive-loading technique employed pre- The effect of the electrically smaller aperture on the directivity is
viously for the dipole antenna. shown in Figure 15 for the uniformly loaded and unloaded spiral. It
was apparent that the directivity for the uniformly loaded spiral
was several dB lower that for the unloaded spiral, and it exhibited
To determine the optimal miniaturization factor, mopt ' for a
an oscillatory behavior. The oscillatory behavior was caused by the
given ka, we follow the same procedure used for the dipole spiral not being able to radiate as effectively (i.e., higher Q) from
antenna. That is, we observed how the Q varies as a function of the the electrically smaller radiation bands. That is, the current was not
effective size kma for several values of ka. The results are shown attenuated sufficiently as it passed through the electrically small
in Figure 11 and, for each case, the spiral Q has been normalized to radiation bands [2]. Therefore, a large amount of current reached
the theoretical Q limit so that all of the curves can be displayed on the end of the spiral arm, where it was reflected and gave rise to the
the same plot. It is apparent that in each case, the Q was minimized oscillations.
when kma ~ 0.57l', implying an optimal miniaturization factor of
mopt ~ 0.57l'/ ka . It is important to note that mopt is larger than the
required m to just fit the radiation band inside the spiral aperture. An obvious solution to this problem is to only load the low-
Such a result is not an unexpected result, because a spiral does not frequency portion of the spiral aperture, or to use a tapered loading
radiate as effectively from the radiation bands located near the profile. In doing so, there will be minimal size reduction of the
aperture edges. That is, if the radiation band is located too close to high-frequency radiation bands. However, it is impossible to
the edge of the aperture, the current does not have sufficient space miniaturize a broadband antenna without affecting some of the
to decay before it reaches the end of the spiral arm. Consequently, higher-frequency components. To demonstrate the improvement
a strong reflection occurs, resulting in a standing-wave-type current obtained by tapering, the inductive loading was exponentially
distribution, instead of the typical decaying current distribution. tapered along the length of the spiral arm, starting with an induc-
Just how close the radiation band can be to the aperture edge tance ofO.OOl).1H and increasing to 5JlH. The effect on the spiral Q
depends upon the growth rate. Thus, for a spiral, mopt will most and broadside directivity are shown in Figure 13 and Figure 15,
respectively. It was clear that the Q for the tapered loading was
likely depend to some extent on the growth rate. Nevertheless,
almost identical to the unloaded spiral at high frequencies (above
estimating mopt in this manner still provides a useful result, since
600 MHz), as desired. Similarly, the directivity for the tapered
it can be used as a starting point in the design process for defining loading was almost identical to the unloaded spiral. Therefore, a
the maximum miniaturization factor that needs to be achieved. tapered loading profile is an attractive (if not necessary) choice for
broadband antenna miniaturization, because it simultaneously low-
ers the Q at low frequencies (below 600 MHz), while minimizing
negative effects of miniaturization at higher frequencies.
4.2 Loading Profile

In designing a miniaturized broadband antenna, it is impor- It is important to note that the tapered loading profile pro-
tant to consider the manner in which the antenna is miniaturized. vides the best overall performance. However, uniform loading
As an example, consider the miniaturization of a spiral antenna by results in the best low-frequency performance, at the expense of
a factor m. This miniaturization factor can be achieved by uni- high-frequency performance. This is illustrated in Figure 15, which
formly loading the spiral such that the phase velocity is reduced by compares the broadside realized gain for a spiral antenna with a
a factor m everywhere along the spiral. On the other hand, the same uniform loading profile and an exponentially tapered loading pro-
miniaturization factor can also be achieved by gradually decreasing file. In each case, the antennas were matched to purely real load
the phase velocity along the spiral using a tapered loading profile. impedances that maximized their overall performance. From Fig-
An obvious question is whether one approach is better than the ure 15, it was evident that below about 600 MHz, the uniformly
other. To answer this question, let us consider the uniform loading loaded spiral had the highest realized gain. As all three antennas
of the six-inch-diameter spiral shown in Figure 12. For a frequency were lossless with almost the same directivity (see Figure 14), this
fo, there is a radiation band the location of which is shown in Fig- higher realized gain can be attributed to improved matching. That
ure 12, prior to miniaturization. It is obvious that the spiral is elec- is, the uniformly loaded spiral had the lowest Q in this region as
trically large enough to naturally establish a radiation band at /0' seen in Figure 13. However, above 600 MHz, the realized gain of
Therefore, its performance at /0 is sufficient, and needs no further both the unloaded and tapered spirals was 5 to 8 dB better, which
was a significant improvement. This improvement was a product
improvement. When the spiral is miniaturized by a factor m using a
of a slightly higher directivity (1-2 dB), and better matching in both
uniform loading profile, the location of the radiation band associ-
cases due to their lower Q. Figure 15 illustrates an important aspect
ated with /0 will then shift inward, as shown in Figure 12. How-
of miniaturizing a broadband antenna that was alluded to earlier in
ever, in doing so, the size of the radiation band is also reduced by a our discussion of broadband impedance matching. That is, the per-
factor of m. Therefore, after miniaturization, the spiral now radiates formance of a broadband antenna can be further improved at fre-
from an electrically smaller aperture at the frequency /0' implying quencies where it is electrically small as long as one is willing to
a higher antenna Q. Thus, while uniform loading can improve per- sacrifice performance at frequencies where it is electrically large.
formance at frequencies where the spiral is electrically small Therefore, there exists an inherent tradeoff in the miniaturization
(ka < 1), it can also lead to worse performance at those frequencies of broadband antennas, and this is a consequence of the funda-
(ka > 1) where the spiral could already form a radiation band. mental limits associated with the radiation Q.

66 IEEEAntennasand Propagation Magazine, Vol. 51, No.4, August 2009


2 r;::::===::::;--.---- ---.---- -,--- ,
- - ka - 0.05
- - - ka - 0.1 :- . .. .... . - - ka - 0.05
1.8 \

1.8 - - - ka - 0.1
··· ······ ka - 0.15 : :
... . ···· ·ka - 0.15
1.6 - - . ka - 0.2 : :" . . .. .. .... .. ..... .
1.6 _ ._ . . ka - 0.2
~

-=-.k_a_-_0_.25-t /
,/ .,
1.4 rL--=-
-=-
-=- _ , : . . ~ :- . - - ka - 0.25
1.4
: .:. / :"
,/
1.2 .. ...:- \ : ; . . 1.2 .
el' .... .... .............: .
a
0.8
0.8

0.6 ·· · ·· · : · ·· ··· ·!·······[·······:· · ·· · · ·:· ·· ··· · t· ···(·· · · t" · · · · ~ · · · · · i··· . 0.6 ..........: ! !" ; .
0.4 · · ··· · ~·· · ·· · ·,·······:······ r· ·· · · : · · · · · · ·(" · · · ··: ···· · · j· · · · · · · ·! · · · · · · · 1· · · · · · 0.4 .... •. . : .•••.•••••••.••• : .•.•.•••••...... -:..••• ..... . . . . . . . !
.
.

0.2 :- , ~ , : ·······r· , , : : . 0.2 .


.
. : ~ .
o °O~----:':;----~----c'=-----'-----_--.J
1 2 3 4 5 6 7 8 9 10 11 12
0.5 1.5 2 2.5
Miniaturization Factor, m

Figure 9. The normalized dipole Q (at a fixed ka) as a function Figure 10. The dipole Q as a function of kma = mka.
of the miniaturization factor, m.

100 r;:::r===:::;----:-----.-----,-7,-r--, 1000 I~n::r===:::::c==========~


90 ------ ~: : ~~ : ! : /. 900
:, : - - L - O I1H/m
........;l" ....: - - - Uniform, L -6 ~m
.. ....... ka - 0.3 : : . (
80 ~ ~: : ~:: ··· ··· ·.. ·T"··············Y'·· ··.. ···r······ / ······· 800 ..... ...:.•... ..: - .- .•Exponential Taper, L - 0.001-6 l1H1m
70 I ~ : j . 700 . . ....·i· .\.··.: 1... ; ~ L : .L .
60 I - ~: 'r :" :' ~ ~ . 600

aI 50
a 40
"1 500

400
30 300

20 200 ~ ~ "': .: : ~ .
10 , : ; .
100 ········:········· .. ·~~:4.:.".::~·,: · ·····j/!.~· ~·, ·~ J::;,/ ~.~. >
~OO 200 300 400 500 600 700 800 900 1000
frequency (MHz)

Figure 11. The spiral antenna's Q as a function of the effective Figure 13. A comparison of the radiation Q for a six-inch-
electrical size kma for fixed values of ka. diameter spiral antenna having uniform and exponential
inductive loading profiles.

IEEE Antennasand Propagation Magazine, Vol. 51, No.4, August 2009 67


5 r;====:::c=============r-:--~ loading makes it possible to maintain the same impedance
--L=O~/m
throughout the entire spiral structure as implied by Equation (23) .
4.5 - - - Uniform, L =5 ~/m
Therefore, ideally, an antenna should be miniaturized using equal
4 _ ._ .. Exponential Taper, L .. 0.001 ~ IlH/m
inductive and capacitive loading.
u . '
~ 3.5 , : : ; : :

.~ 3 ····· ··r ···· ····: ······· ··:····· ····j·········;··


e 5. Conclusion
..
~ 2.5
~
u
...
is
--~--'i---t< tT\L! ,/; \ J In this paper, the fundamental limits on radiation Q, directiv-
B 1.5 .. ......, ;
.
-: :..
.
..\./ .. ity, and impedance matching of antennas were reviewed and dis-
{!.
: ~
cussed, with particular emphasis on the Fano-Bode impedance-
1 . ..; ~ , ..: .. matching limitations associated with the bandpass and high -pass
0.5 .: ! ,. : ! (. .. response. For the high-pass case, we showed how the Fano-Bode
limits restrict the size of a UWB antenna, subject to some desired
oL..-_ - ' - _ - ' - _ - - l_ _" - - _ - - ' - _ - ' - _ - - l_ _"------l pass-band reflection coefficient or realized gain. We then discussed
100 200 300 400 500 600 700 ODD 900 1000
frequency (MHz) how miniaturization techniques can be used with preexisting UWB
antenna designs to approach this limit by minimizing the antenna's
Figure 14. A comparison of the broadside directive gain for a Q. Finally, we outl ined a methodology for miniaturizing a UWB
six-inch-diameter spiral antenna having a uniform and an antenna using a spiral antenna as an example.
exponential inductive loading profile.

5 ,--- .---- -,..-- --r- - , - - --.-- - --.- - --r-- .----,


6. References
. . ...
u
o .
,/ <;.:-:->.-< '. 1. V. H. Rumsey, Frequency Independent Antennas , New York,
iii /. / : Academic Press, 1966 .
~ .5 J .:."!' /- : :.
C / :.
'iii . f ;/
~.. -10
..//.<
I "
: .. .. .. .. . ..:..... .. . , 2. R. H. DuHamel and J. P. Scherer, "Frequency-Independent
Antennas," in R. C. Johnson (ed.) , Antenna Engineering Hand-
.~ I .I book, Third Edition, New York, McGraw-Hill, Inc ., 1993, Chapter

i" /i - ' i i -
14, pp . 18-25 .

3. K. Fujimoto, A. Henderson, K. Hirasawa and J. R. James, Small


-20 .. .. JI :/
..
......: --L"O~/m Antennas, New York, John Wiley & Sons, 1987.
I .I - - - Uniform, L .. 5 IlH/m
/.' : : _ ._ .. Exponential Taper, L .. 0.001 ~ IlH/m 4. H. A. Wheeler, "The Radian Sphere Around a Small Antenna,"
-25
100 200 300 400 500 600 700 BOO 900 1000 Proceedings ofthe I.R.E., 35, August 1959, pp. 1325-13 31.
frequency (MHz)
5. L. J. Chu, "Physical Limitations of Antenna Q," Journal of
Figure 15. A comparison of the broadside realized gain for a Applied Physics , 19, December 1948, pp. 1163-1175.
six-inch-diameter spiral antenna for three different loadings:
unloaded (L =0 pH), uniform (L =5 pH), and exponential 6. H. A. Wheeler, "Fundamental Limitations of Small Antennas,"
(L = 0.001 to 5pH) inductive loading profile . Proceedings ofthe I.R.E., 35, December 1947, pp . 1479-1484.

7. S. A. Schelkunoff, "A Mathematical Theory of Linear Arrays,"


Bell System Tech. J., 22,1943, pp. 80-107.
4.3 Equal LC Loading 8. R. F. Harrington, Time-Harmonic Electromagnetic Fields, New
York, John Wiley & Sons, Inc ., 2001.
Even though a tapered loading profile allev iates many high-
frequency issues, it can cause additional issues when the loading is 9. R. F. Harrington, "On the Gain and Beamwidth of Direct ional
purely inductive or capacitive. Similar to a tapered transmission- Antennas," I.R.E. Transactions on Antennas and Propagation, AP-
line matching sect ion , a tapered loading profile introduces imped- 6, January 1958 , pp . 219 -225 .
ance discontinuities along the length of the spiral arm that in tum
cause reflections. The magnitude of these reflections depends upon 9. R. L. Fante, "M aximum Possible Gain for an Arbitrary Ideal
how large the impedance discontinuity is from one section to the Antenna with Specified Quality Factor," IEEE Transactions on
nex t. Of course, as long as the impedance differences are small, the Antennas and Propagation, AP-40, 12, December 1992, pp . 1586-
reflections due to incremental impedance changes will also be 1588.
small, and will have minimal impact on the antenna's input imped-
ance. However, in theory, it is possible to eliminate reflections that 11. R. L. Fante, " Quality Factor of General Ideal Antennas," IEEE
may occur if both inductive and capacitive load ing are simultane- Transactions on Antennas and Propagation , AP-17 , 2, March
ously used. That is, concurrent use of inductive and capacitive 1969, pp . 151-155 .

68 IEEE Antennas and Propagation Magazine, Vol. 51, No. 4, August 2009
12. R. P. Feynman, R. Leighton, and M. Sands, The Feynmen Lec- 25. H. A. Wheeler, "Wideband Impedance Matching," Wheeler
tures on Physics: Volume 1, Upper Saddle River, NJ, Addison- Labs Tech. Rep. 418, May 1950, available at
Wesley, 1961. http://www.arlassociates.net.

13. R. F. Harrington, "Effect of Antenna Size on Gain, Bandwidth, 26. M. C. Villalobos, H. D. Foltz, 1. S. McLean, and I. S. Gupta,
and Efficiency," J. Res. Nat. Bureau Stand., 64D, January 1960, "Broadband Tuning Limits on UWB Antennas Based on Fano's
pp. 1-12. Formulation," IEEE International Symposium on Antennas and
Propagation Digest, July 2006, pp. 171-174.
14. R. E. Collin and S. Rothschild, "Evaluation of Antenna Q,"
IEEE Transactions on Antennas and Propagation, AP-12, January 27. H. L. ThaI, "New Radiation Q Limits for Spherical Wire
1964, pp. 23-27. Antennas," IEEE Transactions on Antennas and Propagation, AP-
54, 10, October 2006, pp. 2757-2763.
15. J. S. McLean, "A Re-Examination of the Fundamental Limits
on the Radiation Q of Electrically Small Antennas," IEEE Trans- 28. A. R. Lopez, "Fundamental Limitations of Small Antennas:
actions on Antennas and Propagation, AP-44, 5, May 1996, pp. Validation of Wheeler's Formulas," IEEE Antennas and Propaga-
672-676. tion Magazine, 48, 4, August 2006, pp. 28-36.

16. W. Geyi, "Physical Limitations of Antenna," IEEE Transac- 29. S. A. Schelkunoff, "Theory of Antennas of Arbitrary Size and
tions on Antennas and Propagation, AP-51, 8, August 2003, pp. Shape," Proceedings ofthe IR.E., 29, 9, September 1941, pp. 493-
2116-2123. 521.

17. H. W. Bode, Network Analysis and Feedback Amplifier 30. J. D. Kraus and R. J. Marhefka, Antennas, Third Edition, New
Design, New York, D. Van Nostrand Co., 1945. York, McGraw-Hill, Inc., 2001.

18. R. M. Fano, "Limitations on the Broadband Matching of Arbi- 31. G. J. Burke, E. K. Miller, and A. J. Poggio, "The Numerical
trary Impedances," Journal of the Franklin Institute, 249, 1, Janu- Electromagnetics Code (NEC) - A Brief History," IEEE Interna-
ary and February 1960, pp. 57-83, 139-154. tional Symposium on Antennas and Propagation Digest, June
2004,3,pp.2871-2874.
19. R. M. Fano, Theoretical Limitations on the Broadband
Matching of Arbitrary Impedances, PhD dissertation, Massachu- 32. A. D. Yaghjian and S. R. Best, "Impedance, Bandwidth, and Q
setts Institute of Technology, 1947. of Antennas," IEEE Transactions on Antennas and Propagation,
AP-53, 4, April 2005, pp. 1298-1324.
20. A. R. Lopez, "Review of Narrowband Impedance-Matching
Limitations," IEEE Antennas and Propagation Magazine, 46, 4, 33. M. Lee, B. A. Kramer, C.-C. Chen, and J. L. Volakis, "Distrib-
August 2004, pp. 88-90. uted Lumped Loads and Lossy Transmission Line Model for
Wideband Spiral Antenna Miniaturization and Characterization,"
21. D. M. Pozar, Microwave Engineering, Second Edition, New IEEE Transactions on Antennas and Propagation, AP-55, 10,
York, John Wiley & Sons, Inc., 1998. October 2007, pp. 2671-2678.

22. R. C. Hansen, Electrically Small, SuperDirective, and. Super- 34. B. A. Kramer, C.-C. Chen, and J. L. Volakis, "Size Reduction
conducting Antennas, New York, John Wiley & Sons, Inc., 2006. of a Low-Profile Spiral Antenna Using Inductive and Dielectric
Loading," IEEE Antennas and Wireless Propagation Letters, 7,
23. H. A. Wheeler, "The Wide-Band Matching Area for a Small 2008, pp. 22-25.
Antenna," IEEE Transactions on Antennas and Propagation, AP-
31,2, March 1983, pp. 364-367. 35. R. Bawer and J. J. Wolfe, "The Spiral Antenna," I.R.E. Inter-
national Convention Record, March 1960. @
24. G. L. Matthaei, L. Young, and E. M. T. Jones, Microwave Fil-
ters, Impedance-Matching Networks and Coupling Structures,
New York, McGraw-Hill Book Company, Inc., 1964.

IEEE Antennasand Propagation Magazine, Vol. 51, No.4, August 2009 69

You might also like