242 Electromagnetic Waves & Antennas – S. J.

Orfanidis

Rectangular waveguides are used routinely to transfer large amounts of microwave

8 power at frequencies greater than 3 GHz. For example at 5 GHz, the transmitted power
might be one megawatt and the attenuation only 4 dB/100 m.
Optical fibers operate at optical and infrared frequencies, allowing a very wide band-

Waveguides width. Their losses are very low, typically, 0.2 dB/km. The transmitted power is of the
order of milliwatts.

8.1 Longitudinal-Transverse Decompositions

In a waveguiding system, we are looking for solutions of Maxwell’s equations that are
propagating along the guiding direction (the z direction) and are confined in the near
vicinity of the guiding structure. Thus, the electric and magnetic fields are assumed to
have the form:
Waveguides are used to transfer electromagnetic power efficiently from one point in
space to another. Some common guiding structures are shown in the figure below. E(x, y, z, t)= E(x, y)ejωt−jβz
(8.1.1)
These include the typical coaxial cable, the two-wire and mictrostrip transmission lines, H(x, y, z, t)= H(x, y)ejωt−jβz
hollow conducting waveguides, and optical fibers.
where β is the propagation wavenumber along the guide direction. The corresponding
In practice, the choice of structure is dictated by: (a) the desired operating frequency
wavelength, called the guide wavelength, is denoted by λg = 2π/β.
band, (b) the amount of power to be transferred, and (c) the amount of transmission
The precise relationship between ω and β depends on the type of waveguiding struc-
losses that can be tolerated.
ture and the particular propagating mode. Because the fields are confined in the trans-
verse directions (the x, y directions,) they cannot be uniform (except in very simple
structures) and will have a non-trivial dependence on the transverse coordinates x and
y. Next, we derive the equations for the phasor amplitudes E(x, y) and H(x, y).
Because of the preferential role played by the guiding direction z, it proves con-
venient to decompose Maxwell’s equations into components that are longitudinal, that
is, along the z-direction, and components that are transverse, along the x, y directions.
Thus, we decompose:

Fig. 8.0.1 Typical waveguiding structures.

E(x, y)= x̂ Ex (x, y)+ŷ Ey (x, y) + ẑ Ez (x, y) ≡ ET (x, y)+ẑ Ez (x, y) (8.1.2)

 
 
Coaxial cables are widely used to connect RF components. Their operation is practi- transverse longitudinal
cal for frequencies below 3 GHz. Above that the losses are too excessive. For example,
the attenuation might be 3 dB per 100 m at 100 MHz, but 10 dB/100 m at 1 GHz, and In a similar fashion we may decompose the gradient operator:
50 dB/100 m at 10 GHz. Their power rating is typically of the order of one kilowatt at
100 MHz, but only 200 W at 2 GHz, being limited primarily because of the heating of ∇ = x̂ ∂x + ŷ ∂y + ẑ ∂z = ∇ T + ẑ ∂z = ∇ T − jβ ẑ (8.1.3)

 
the coaxial conductors and of the dielectric between the conductors (dielectric voltage transverse
breakdown is usually a secondary factor.) where we made the replacement ∂z → −jβ because of the assumed z-dependence. In-
Another issue is the single-mode operation of the line. At higher frequencies, in order troducing these decompositions into the source-free Maxwell’s equations we have:
to prevent higher modes from being launched, the diameters of the coaxial conductors
must be reduced, diminishing the amount of power that can be transmitted.
∇ × E = −jωµH ∇T − jβẑ)×(ET + ẑ Ez )= −jωµ(HT + ẑ Hz )
(∇
Two-wire lines are not used at microwave frequencies because they are not shielded
and can radiate. One typical use is for connecting indoor antennas to TV sets. Microstrip ∇ × H = jωE ∇T − jβẑ)×(HT + ẑ Hz )= jω(ET + ẑ Ez )
(∇
lines are used widely in microwave integrated circuits. ⇒ (8.1.4)
∇·E=0 ∇T − jβẑ)·(ET + ẑ Ez )= 0
(∇

∇·H=0 ∇T − jβẑ)·(HT + ẑ Hz )= 0
(∇
8.1. Longitudinal-Transverse Decompositions 243 244 Electromagnetic Waves & Antennas – S. J. Orfanidis

where , µ denote the permittivities of the medium in which the fields propagate, for The solution of this system is:
example, the medium between the coaxial conductors in a coaxial cable, or the medium
within the hollow rectangular waveguide. This medium is assumed to be lossless for jβ jωµ
ẑ × ET = − ẑ × ∇ T Ez − ∇ T Hz
now. k2c k2c
(8.1.7)
We note that ẑ · ẑ = 1, ẑ × ẑ = 0, ẑ · ET = 0, ẑ · ∇ T Ez = 0 and that ẑ × ET and jω jβ
HT = − ẑ × ∇ T Ez − 2 ∇ T Hz
ẑ × ∇ T Ez are transverse while ∇ T × ET is longitudinal. Indeed, we have: k2c kc

ẑ × ET = ẑ × (x̂ Ex + ŷ Ey )= ŷ Ex − x̂ Ey where we defined the so-called cutoff wavenumber kc by:

∇ T × ET = (x̂ ∂x + ŷ ∂y )×(x̂ Ex + ŷ Ey )= ẑ(∂x Ey − ∂y Ex )
ω2
kc2 = ω2 µ − β2 = − β2 = k2 − β2 (cutoff wavenumber) (8.1.8)
Using these properties and equating longitudinal and transverse parts in the two c2
sides of Eq. (8.1.4), we obtain the equivalent set of Maxwell equations: √
The quantity k = ω/c = ω µ is the wavenumber a uniform plane wave would
have in the propagation medium , µ.
∇ T Ez × ẑ − jβ ẑ × ET = −jωµHT
Although k2c stands for the difference ω2 µ − β2 , it turns out that the boundary
∇ T Hz × ẑ − jβ ẑ × HT = jωET
conditions for each waveguide type force k2c to take on certain values, which can be
∇ T × ET + jωµ ẑ Hz = 0
(8.1.5) positive, negative, or zero, and characterize the propagating modes. For example, in a
∇ T × HT − jω ẑ Ez = 0 dielectric waveguide k2c is positive inside the guide and negative outside it; in a hollow
∇ T · ET − jβEz = 0 conducting waveguide k2c takes on certain quantized positive values; in a TEM line, k2c
∇ T · HT − jβHz = 0 is zero. Some related definitions are the cutoff frequency and the cutoff wavelength
Depending on whether both, one, or none of the longitudinal components are zero, defined as follows:
we may classify the solutions as transverse electric and magnetic (TEM), transverse elec-
2π
tric (TE), transverse magnetic (TM), or hybrid: ωc = ckc , λc = (cutoff frequency and wavelength) (8.1.9)
kc
Ez = 0, Hz = 0, TEM modes We can then express β in terms of ω and ωc , or ω in terms of β and ωc . Taking
Ez = 0, Hz = 0, TE or H modes the positive square roots of Eq. (8.1.8), we have:
Ez = 0, Hz = 0, TM or E modes

Ez = 0, Hz = 0, hybrid or HE or EH modes  
1 ω ωc2
β= ω2 − ω2c = 1− and ω = ω2c + β2 c2 (8.1.10)
c c ω2
In the case of TEM modes, which are the dominant modes in two-conductor trans-
mission lines such as the coaxial cable, the fields are purely transverse and the solution Often, Eq. (8.1.10) is expressed in terms of the wavelengths λ = 2π/k = 2πc/ω,
of Eq. (8.1.5) reduces to an equivalent two-dimensional electrostatic problem. We will λc = 2π/kc , and λg = 2π/β. It follows from k2 = k2c + β2 that
discuss this case later on.
1 1 1 λ
In all other cases, at least one of the longitudinal fields Ez , Hz is non-zero. It is then = + ⇒ λg =  (8.1.11)
possible to express the transverse field components ET , HT in terms of the longitudinal λ2 λ2c λ2g λ2
1−
ones, Ez , Hz . λ2c
Forming the cross-product of the second of equations (8.1.5) with ẑ and using the Note that λ is related to the free-space wavelength λ0 = 2πc0 /ω = c0 /f by the
BAC-CAB vector identity, ẑ × (ẑ × HT )= ẑ(ẑ · HT )−HT (ẑ · ẑ)= −HT , and similarly, refractive index of the dielectric material λ = λ0 /n.
∇T Hz × ẑ)= ∇ T Hz , we obtain:
ẑ × (∇ It is convenient at this point to introduce the transverse impedances for the TE and
TM modes by the definitions:
∇ T Hz + jβHT = jω ẑ × ET

Thus, the first two of (8.1.5) may be thought of as a linear system of two equations ωµ ω β βc
in the two unknowns ẑ × ET and HT , that is, ηTE = =η , ηTM = =η (TE and TM impedances) (8.1.12)
β βc ω ω
β ẑ × ET − ωµHT = jẑ × ∇ T Ez 
(8.1.6) where the medium impedance is η = µ/, so that η/c = µ and ηc = 1/. We note the
∇T Hz
ω ẑ × ET − βHT = −j∇ properties:
8.1. Longitudinal-Transverse Decompositions 245 246 Electromagnetic Waves & Antennas – S. J. Orfanidis

ηTE ω2 ∇2T = ∇ T · ∇ T = ∂2x + ∂2y (8.1.18)

ηTE ηTM = η2 , = 2 2 (8.1.13)
ηTM β c 2
and we used the vectorial identities ∇ T × ∇ T Ez = 0, ∇ T × (ẑ × ∇ T Hz )= ẑ ∇T Hz , and
 ∇ T · (ẑ × ∇ T Hz )= 0.
Because βc/ω = 1 − ω2c /ω2 , we can write also:
It follows that in order to satisfy all of the last four of Maxwell’s equations (8.1.5), it
 is necessary that the longitudinal fields Ez (x, y), Hz (x, y) satisfy the two-dimensional
η ω2c Helmholtz equations:
ηTE =  , ηTM = η 1− (8.1.14)
ω2 ω2
1 − 2c ∇2T Ez + k2c Ez = 0
ω
(Helmholtz equations) (8.1.19)
With these definitions, we may rewrite Eq. (8.1.7) as follows: ∇2T Hz + k2c Hz = 0

jβ  These equations are to be solved subject to the appropriate boundary conditions for
ẑ × ET = − ẑ × ∇ T Ez + ηTE∇ T Hz each waveguide type. Once, the fields Ez , Hz are known, the transverse fields ET , HT are
k2c
(8.1.15) computed from Eq. (8.1.16), resulting in a complete solution of Maxwell’s equations for
jβ  1 the guiding structure. To get the full x, y, z, t dependence of the propagating fields, the
HT = − ẑ × ∇ T Ez + ∇ T Hz
k2c ηTM above solutions must be multiplied by the factor ejωt−jβz .
Using the result ẑ × (ẑ × ET )= −ET , we solve for ET and HT : The cross-sections of practical waveguiding systems have either cartesian or cylin-
drical symmetry, such as the rectangular waveguide or the coaxial cable. Below, we
jβ  summarize the form of the above solutions in the two types of coordinate systems.
ET = − ∇ T Ez − ηTE ẑ × ∇ T Hz
k2c
(transverse fields) (8.1.16)
jβ  1 Cartesian Coordinates
HT = − ∇ T Hz + ẑ × ∇ T Ez
k2c ηTM
The cartesian component version of Eqs. (8.1.16) and (8.1.19) is straightforward. Using
An alternative and useful way of writing these equations is to form the following the identity ẑ × ∇ T Hz = ŷ ∂x Hz − x̂ ∂y Hz , we obtain for the longitudinal components:
linear combinations, which are equivalent to Eq. (8.1.6):
(∂2x + ∂2y )Ez + k2c Ez = 0
1 j (8.1.20)
HT − ẑ × ET = ∇ T Hz (∂2x + ∂2y )Hz + k2c Hz = 0
ηTM β
(8.1.17)
j Eq. (8.1.16) becomes for the transverse components:
ET − ηTE HT × ẑ = ∇ T Ez
β

So far we only used the first two of Maxwell’s equations (8.1.5) and expressed ET , HT jβ  jβ  1
Ex = − ∂x Ez + ηTE ∂y Hz Hx = − ∂x Hz − ∂y Ez
in terms of Ez , Hz . Using (8.1.16), it is easily shown that the left-hand sides of the k2c k2c ηTM
remaining four of Eqs. (8.1.5) take the forms: , (8.1.21)
jβ  jβ  1
Ey = − 2 ∂y Ez − ηTE ∂x Hz Hy = − 2 ∂y Hz + ∂x Ez
jωµ  2 kc kc ηTM
∇ T × ET + jωµ ẑ Hz = ẑ ∇T Hz + k2c Hz
k2c
jω  Cylindrical Coordinates
∇ T × HT − jω ẑ Ez = − 2 ẑ ∇2T Ez + k2c Ez
kc
The relationship between cartesian and cylindrical coordinates is shown in Fig. 8.1.1.
jβ  2
∇ T · ET − jβEz = − 2 ∇T Ez + k2c Ez From the triangle in the figure, we have x = ρ cos φ and y = ρ sin φ. The transverse
kc
gradient and Laplace operator are in cylindrical coordinates:
jβ 
∇ T · HT − jβHz = − 2 ∇2T Hz + k2c Hz
kc ∂ 1 ∂ 1 ∂ ∂ 1 ∂2
∇ T = ρ̂
ρ + φ̂
φ , ∇ 2T = ρ + (8.1.22)
∂ρ ρ ∂φ ρ ∂ρ ∂ρ ρ2 ∂φ2
where ∇2T is the two-dimensional Laplacian operator:
The Helmholtz equations (8.1.19) now read:
8.2. Power Transfer and Attenuation 247 248 Electromagnetic Waves & Antennas – S. J. Orfanidis

1
PT = Pz dS , where Pz = Re(E × H∗ )·ẑ (8.2.1)
S 2

It is easily verified that only the transverse components of the fields contribute to
the power flow, that is, Pz can be written in the form:

1
Pz = Re(ET × H∗
T )·ẑ (8.2.2)
Fig. 8.1.1 Cylindrical coordinates. 2
For waveguides with conducting walls, the transmission losses are due primarily to
ohmic losses in (a) the conductors and (b) the dielectric medium filling the space between
the conductors and in which the fields propagate. In dielectric waveguides the losses
1 ∂ ∂Ez 1 ∂2 Ez are due to absorption and scattering by imperfections.
ρ + + k2c Ez = 0
ρ ∂ρ ∂ρ ρ2 ∂φ2 The transmission losses can be quantified by replacing the propagation wavenumber
(8.1.23)
β by its complex-valued version βc = β − jα, where α is the attenuation constant. The
1 ∂ ∂Hz 1 ∂2 Hz 2
ρ + + kc Hz = 0 z-dependence of all the field components is replaced by:
ρ ∂ρ ∂ρ ρ2 ∂φ2

e−jβz → e−jβc z = e−(α+jβ)z = e−αz e−jβz (8.2.3)

Noting that ẑ × ρ̂
ρ = φ̂
φ and ẑ × φ̂
φ = −ρ̂
ρ, we obtain:
The quantity α is the sum of the attenuation constants arising from the various loss
1
ẑ × ∇ T Hz = φ̂
φ(∂ρ Hz )−ρ̂
ρ (∂φ Hz ) mechanisms. For example, if αd and αc are the attenuations due to the ohmic losses in
ρ
the dielectric and in the conducting walls, then
The decomposition of a transverse vector is ET = ρ̂
ρ Eρ + φ̂
φ Eφ . The cylindrical
α = αd + αc (8.2.4)
coordinates version of (8.1.16) are:
The ohmic losses in the dielectric can be characterized either by its loss tangent, say
tan δ, or by its conductivity σd —the two being related by σd = ω tan δ. The effective
jβ  1 jβ  1 dielectric constant of the medium is then (ω)=  − jσd /ω = (1 − j tan δ). The
Eρ = − ∂ρ Ez − ηTE ∂φ Hz Hρ = − ∂ρ Hz + ∂φ Ez
k2c ρ k2c ηTM ρ corresponding complex-valued wavenumber βc is obtained by the replacement:
, (8.1.24)
jβ  1 jβ  1 1  
Eφ = − ∂φ Ez + ηTE ∂ρ Hz Hφ = − ∂φ Hz − ∂ρ Ez
k2c ρ k2c ρ ηTM β = ω2 µ − k2c → βc = ω2 µ(ω)−k2c

For either coordinate system, the equations for HT may be obtained from those of For weakly conducting dielectrics, we may make the approximation:
ET by a so-called duality transformation, that is, making the substitutions:  
 σd ωµσd 1 ωµ
βc = ω2 µ 1−j − k2c = β2 − jωµσd = β 1−j  β − j σd
ω β2 2 β
E → H, H → −E ,  → µ, µ→ (duality transformation) (8.1.25)
Recalling the definition ηTE = ωµ/β, we obtain for the attenuation constant:
−1
These imply that η → η−1 and ηTE → ηTM . Duality is discussed in greater detail in
Sec. 16.2.
1 1 ω2 ω tan δ
αd = σd ηTE = tan δ =  (dielectric losses) (8.2.5)
2 2 βc2 2c 1 − ω2c /ω2
8.2 Power Transfer and Attenuation
which is similar to Eq. (2.7.2), but with the replacement ηd → ηTE .
With the field solutions at hand, one can determine the amount of power transmitted
The conductor losses are more complicated to calculate. In practice, the following
along the guide, as well as the transmission losses. The total power carried by the fields
approximate procedure is adequate. First, the fields are determined on the assumption
along the guide direction is obtained by integrating the z-component of the Poynting
that the conductors are perfect.
vector over the cross-sectional area of the guide:
8.3. TEM, TE, and TM modes 249 250 Electromagnetic Waves & Antennas – S. J. Orfanidis

Second, the magnetic fields on the conductor surfaces are determined and the corre- One common property of all three types of modes is that the transverse fields ET , HT
sponding induced surface currents are calculated by Js = n̂ × H, where n̂ is the outward are related to each other in the same way as in the case of uniform plane waves propagat-
normal to the conductor. ing in the z-direction, that is, they are perpendicular to each other, their cross-product
Third, the ohmic losses per unit conductor area are calculated by Eq. (2.8.7). Figure points in the z-direction, and they satisfy:
8.2.1 shows such an infinitesimal conductor area dA = dl dz, where dl is along the
cross-sectional periphery of the conductor. Applying Eq. (2.8.7) to this area, we have: 1
HT = ẑ × ET (8.3.1)
ηT
dPloss dPloss 1
= = Rs |Js |2 (8.2.6) where ηT is the transverse impedance of the particular mode type, that is, η, ηTE , ηTM
dA dldz 2
where Rs is the surface resistance of the conductor given by Eq. (2.8.4), in the TEM, TE, and TM cases.
 Because of Eq. (8.3.1), the power flow per unit cross-sectional area described by the
ωµ ω 1 2 Poynting vector Pz of Eq. (8.2.2) takes the simple form in all three cases:
Rs = =η = δωµ , δ= = skin depth (8.2.7)
2σ 2σ 2 ωµσ
1 1 1
Integrating Eq. (8.2.6) around the periphery of the conductor gives the power loss per Pz = Re(ET × H∗
T )·ẑ = |ET |2 = ηT |HT |2 (8.3.2)
unit z-length due to that conductor. Adding similar terms for all the other conductors 2 2ηT 2
gives the total power loss per unit z-length:
  TEM modes
dPloss 1 1
Ploss = = Rs |Js |2 dl + Rs |Js |2 dl (8.2.8) In TEM modes, both Ez and Hz vanish, and the fields are fully transverse. One can set
dz Ca 2 Cb 2
Ez = Hz = 0 in Maxwell equations (8.1.5), or equivalently in (8.1.16), or in (8.1.17).
From any point view, one obtains the condition k2c = 0, or ω = βc. For example, if
the right-hand sides of Eq. (8.1.17) vanish, the consistency of the system requires that
ηTE = ηTM , which by virtue of Eq. (8.1.13) implies ω = βc. It also implies that ηTE , ηTM
must both be equal to the medium impedance η. Thus, the electric and magnetic fields
satisfy:

1
HT = ẑ × ET (8.3.3)
η

These are the same as in the case of a uniform plane wave, except here the fields
are not uniform and may have a non-trivial x, y dependence. The electric field ET is
Fig. 8.2.1 Conductor surface absorbs power from the propagating fields. determined from the rest of Maxwell’s equations (8.1.5), which read:

where Ca and Cb indicate the peripheries of the conductors. Finally, the corresponding
∇ T × ET = 0
attenuation coefficient is calculated from Eq. (2.6.22):
(8.3.4)
∇ T · ET = 0
Ploss
αc = (conductor losses) (8.2.9) These are recognized as the field equations of an equivalent two-dimensional elec-
2PT
trostatic problem. Once this electrostatic solution is found, ET (x, y), the magnetic field
Equations (8.2.1)–(8.2.9) provide a systematic methodology by which to calculate the
is constructed from Eq. (8.3.3). The time-varying propagating fields will be given by
transmitted power and attenuation losses in waveguides. We will apply it to several
Eq. (8.1.1), with ω = βc. (For backward moving fields, replace β by −β.)
examples later on.
We explore this electrostatic point of view further in Sec. 9.1 and discuss the cases
of the coaxial, two-wire, and strip lines. Because of the relationship between ET and HT ,
8.3 TEM, TE, and TM modes the Poynting vector Pz of Eq. (8.2.2) will be:

1 1 1
The general solution described by Eqs. (8.1.16) and (8.1.19) is a hybrid solution with non- Pz = Re(ET × H∗
T )·ẑ = |ET |2 = η|HT |2 (8.3.5)
2 2η 2
zero Ez and Hz components. Here, we look at the specialized forms of these equations
in the cases of TEM, TE, and TM modes.
8.3. TEM, TE, and TM modes 251 252 Electromagnetic Waves & Antennas – S. J. Orfanidis

TE modes
∇2T Ez + k2c Ez = 0
TE modes are characterized by the conditions Ez = 0 and Hz = 0. It follows from the
second of Eqs. (8.1.17) that ET is completely determined from HT , that is, ET = ηTE HT ×ẑ. jβ
ET = − ∇ T Ez
The field HT is determined from the second of (8.1.16). Thus, all field components k2c (TM modes) (8.3.10)
for TE modes are obtained from the equations:
1
HT = ẑ × ET
ηTM
∇2T Hz + k2c Hz = 0
Again, the relationship of ET and HT is identical to that of uniform plane waves
jβ
HT = − ∇ T Hz (TE modes) (8.3.6) propagating in the z-direction, but the wave impedance is now ηTM . The Poynting vector
k2c takes the form:
ET = ηTE HT × ẑ
1 1 1 β2
Pz = Re(ET × H∗
T )·ẑ = |ET |2 = ∇T Ez |2
|∇ (8.3.11)
The relationship of ET and HT is identical to that of uniform plane waves propagating 2 2ηTM 2ηTM k4c
in the z-direction, except the wave impedance is replaced by ηTE . The Poynting vector
of Eq. (8.2.2) then takes the form: 8.4 Rectangular Waveguides
Next, we discuss in detail the case of a rectangular hollow waveguide with conducting
1 1 1 1 β2
Pz = Re(ET × H∗
T )·ẑ = ∇T Hz |2
|ET |2 = ηTE |HT |2 = ηTE 4 |∇ (8.3.7) walls, as shown in Fig. 8.4.1. Without loss of generality, we may assume that the lengths
2 2ηTE 2 2 kc
a, b of the inner sides satisfy b ≤ a. The guide is typically filled with air, but any other
The cartesian coordinate version of Eq. (8.3.6) is: dielectric material , µ may be assumed.

(∂2x + ∂y2 )Hz + k2c Hz = 0

jβ jβ
Hx = − ∂x Hz , Hy = − 2 ∂y Hz (8.3.8)
k2c kc
Ex = ηTE Hy , Ey = −ηTE Hx

And, the cylindrical coordinate version:

Fig. 8.4.1 Rectangular waveguide.
1 ∂ ∂Hz 1 ∂2 Hz 2
ρ + + kc Hz = 0
ρ ∂ρ ∂ρ ρ2 ∂φ2 The simplest and dominant propagation mode is the so-called TE10 mode and de-
jβ ∂Hz jβ 1 ∂Hz (8.3.9) pends only on the x-coordinate (of the longest side.) Therefore, we begin by looking
Hρ = − 2 , Hφ = − 2 for solutions of Eq. (8.3.8) that depend only on x. In this case, the Helmholtz equation
kc ∂ρ kc ρ ∂φ
Eρ = ηTE Hφ , Eφ = −ηTE Hρ reduces to:

∂2x Hz (x)+k2c Hz (x)= 0

where we used HT × ẑ = (ρ̂
ρ Hρ + φ̂
φ Hφ )×ẑ = −φ̂
φ Hρ + ρ̂
ρ Hφ .
The most general solution is a linear combination of cos kc x and sin kc x. However,
TM modes only the former will satisfy the boundary conditions. Therefore, the solution is:

TM modes have Hz = 0 and Ez = 0. It follows from the first of Eqs. (8.1.17) that HT is
−1 Hz (x)= H0 cos kc x (8.4.1)
completely determined from ET , that is, HT = ηTM ẑ × ET . The field ET is determined
from the first of (8.1.16), so that all field components for TM modes are obtained from where H0 is a (complex-valued) constant. Because there is no y-dependence, it follows
the following equations, which are dual to the TE equations (8.3.6): from Eq. (8.3.8) that ∂y Hz = 0, and hence Hy = 0 and Ex = 0. It also follows that:

jβ jβ jβ
Hx (x)= − ∂x Hz = − 2 (−kc )H0 sin kc x = H0 sin kc x ≡ H1 sin kc x
k2c kc kc
8.4. Rectangular Waveguides 253 254 Electromagnetic Waves & Antennas – S. J. Orfanidis

Then, the corresponding electric field will be:

jβ
Ey (x)= −ηTE Hx (x)= −ηTE H0 sin kc x ≡ E0 sin kc x
kc

where we defined the constants:

jβ Fig. 8.4.2 Electric field inside a rectangular waveguide.

H1 = H0
kc
(8.4.2)
jβ ω 8.5 Higher TE and TM modes
E0 = −ηTE H1 = −ηTE H0 = −jη H0
kc ωc
where we used ηTE = ηω/βc. In summary, the non-zero field components are: To construct higher modes, we look for solutions of the Helmholtz equation that are
factorable in their x and y dependence:

Hz (x, y)= F(x)G(y)

Hz (x)= H0 cos kc x Hz (x, y, z, t)= H0 cos kc x ejωt−jβz
jωt−jβz
Hx (x)= H1 sin kc x ⇒ Hx (x, y, z, t)= H1 sin kc x e (8.4.3) Then, Eq. (8.3.8) becomes:
Ey (x)= E0 sin kc x jωt−jβz
Ey (x, y, z, t)= E0 sin kc x e
F (x) G (y)
F (x)G(y)+F(x)G (y)+k2c F(x)G(y)= 0 ⇒ + + k2c = 0 (8.5.1)
Assuming perfectly conducting walls, the boundary conditions require that there be F(x) G(y)
no tangential electric field at any of the wall sides. Because the electric field is in the
Because these must be valid for all x, y (inside the guide), the F- and G-terms must
y-direction, it is normal to the top and bottom sides. But, it is parallel to the left and be constants, independent of x and y. Thus, we write:
right sides. On the left side, x = 0, Ey (x) vanishes because sin kc x does. On the right
side, x = a, the boundary condition requires: F (x) G (y)
= −k2x , = −k2y or
F(x) G(y)
Ey (a)= E0 sin kc a = 0 ⇒ sin kc a = 0
F (x)+k2x F(x)= 0 , G (y)+k2y G(y)= 0 (8.5.2)
This requires that kc a be an integral multiple of π: 2 2
where the constants kx and ky are constrained from Eq. (8.5.1) to satisfy:
nπ
kc a = nπ ⇒ kc = (8.4.4)
a k2c = k2x + k2y (8.5.3)
These are the so-called TEn0 modes. The corresponding cutoff frequency ωc = ckc , The most general solutions of (8.5.2) that will satisfy the TE boundary conditions are
fc = ωc /2π, and wavelength λc = 2π/kc = c/fc are: cos kx x and cos ky y. Thus, the longitudinal magnetic field will be:

cnπ cn 2a Hz (x, y)= H0 cos kx x cos ky y (TEnm modes) (8.5.4)

ωc = , fc = , λc = (TEn0 modes) (8.4.5)
a 2a n
It then follows from the rest of the equations (8.3.8) that:
The dominant mode is the one with the lowest cutoff frequency or the longest cutoff
wavelength, that is, the mode TE10 having n = 1. It has:
Hx (x, y) = H1 sin kx x cos ky y Ex (x, y) = E1 cos kx x sin ky y
π cπ c (8.5.5)
kc = , ωc = , fc = , λc = 2a (TE10 mode) (8.4.6) Hy (x, y) = H2 cos kx x sin ky y Ey (x, y) = E2 sin kx x cos ky y
a a 2a

Fig. 8.4.2 depicts the electric field Ey (x)= E0 sin kc x = E0 sin(πx/a) of this mode where we defined the constants:
as a function of x.
jβkx jβky
H1 = H0 , H2 = H0
k2c k2c
ωky ωkx
E1 = ηTE H2 = jη H0 , E2 = −ηTE H1 = −jη H0
ωc kc ωc kc
8.5. Higher TE and TM modes 255 256 Electromagnetic Waves & Antennas – S. J. Orfanidis

The boundary conditions are that Ey vanish on the right wall, x = a, and that Ex 8.6 Operating Bandwidth
vanish on the top wall, y = b, that is,
All waveguiding systems are operated in a frequency range that ensures that only the
Ey (a, y)= E0y sin kx a cos ky y = 0 , Ex (x, b)= E0x cos kx x sin ky b = 0 lowest mode can propagate. If several modes can propagate simultaneously,† one has
no control over which modes will actually be carrying the transmitted signal. This may
The conditions require that kx a and ky b be integral multiples of π: cause undue amounts of dispersion, distortion, and erratic operation.
A mode with cutoff frequency ωc will propagate only if its frequency is ω ≥ ωc ,
nπ mπ or λ < λc . If ω < ωc , the wave will attenuate exponentially along the guide direction.
kx a = nπ , ky b = mπ ⇒ kx = , ky = (8.5.6)
a b This follows from the ω, β relationship (8.1.10):
These
 correspond to the TEnm modes. Thus, the cutoff wavenumbers of these modes ω2 − ω2c
kc = k2x + k2y take on the quantized values: ω2 = ω2c + β2 c2 ⇒ β2 =
c2
 2  2
nπ mπ If ω ≥ ωc , the wavenumber β is real-valued and the wave will propagate. But if
kc = + (TEnm modes) (8.5.7) ω < ωc , β becomes imaginary, say, β = −jα, and the wave will attenuate in the z-
a b
direction, with a penetration depth δ = 1/α:
The cutoff frequencies fnm = ωc /2π = ckc /2π and wavelengths λnm = c/fnm are:
 e−jβz = e−αz
2  2
n m 1
fnm = c + , λnm =  2  2 (8.5.8) If the frequency ω is greater than the cutoff frequencies of several modes, then all
2a 2b n m
+ of these modes can propagate. Conversely, if ω is less than all cutoff frequencies, then
2a 2b
none of the modes can propagate.
The TE0m modes are similar to the TEn0 modes, but with x and a replaced by y and If we arrange the cutoff frequencies in increasing order, ωc1 < ωc2 < ωc3 < · · · ,
b. The family of TM modes can also be constructed in a similar fashion from Eq. (8.3.10). then, to ensure single-mode operation, the frequency must be restricted to the interval
Assuming Ez (x, y)= F(x)G(y), we obtain the same equations (8.5.2). Because Ez ωc1 < ω < ωc2 , so that only the lowest mode will propagate. This interval defines the
is parallel to all walls, we must now choose the solutions sin kx and sin ky y. Thus, the operating bandwidth of the guide.
longitudinal electric fields is: These remarks apply to all waveguiding systems, not just hollow conducting wave-
guides. For example, in coaxial cables the lowest mode is the TEM mode having no cutoff
Ez (x, y)= E0 sin kx x sin ky y (TMnm modes) (8.5.9) frequency, ωc1 = 0. However, TE and TM modes with non-zero cutoff frequencies do
exist and place an upper limit on the usable bandwidth of the TEM mode. Similarly, in
The rest of the field components can be worked out from Eq. (8.3.10) and one finds
optical fibers, the lowest mode has no cutoff, and the single-mode bandwidth is deter-
that they are given by the same expressions as (8.5.5), except now the constants are
mined by the next cutoff frequency.
determined in terms of E0 :
In rectangular waveguides the smallest cutoff frequencies are f10 = c/2a, f20 =
jβkx jβky c/a = 2f10 , and f01 = c/2b. Because we assumed that b ≤ a, it follows that always
E1 = − E0 , E2 = − E0 f10 ≤ f01 . If b ≤ a/2, then 1/a ≤ 1/2b and therefore, f20 ≤ f01 , so that the two lowest
k2c k2c
cutoff frequencies are f10 and f20 .
1 jωky 1 1 jωkx 1
H1 = − E2 = E0 , H2 = E1 = − H0 On the other hand, if a/2 ≤ b ≤ a, then f01 ≤ f20 and the two smallest frequencies
ηTM ωc kc η ηTM ωc kc η
are f10 and f01 (except when b = a, in which case f01 = f10 and the smallest frequencies
where we used ηTM = ηβc/ω. The boundary conditions on Ex , Ey are the same as are f10 and f20 .) The two cases b ≤ a/2 and b ≥ a/2 are depicted in Fig. 8.6.1.
before, and in addition, we must require that Ez vanish on all walls. It is evident from this figure that in order to achieve the widest possible usable
These conditions imply that kx , ky will be given by Eq. (8.5.6), except both n and m bandwidth for the TE10 mode, the guide dimensions must satisfy b ≤ a/2 so that the
must be non-zero (otherwise Ez would vanish identically.) Thus, the cutoff frequencies bandwidth is the interval [fc , 2fc ], where fc = f10 = c/2a. In terms of the wavelength
and wavelengths are the same as in Eq. (8.5.8). λ = c/f , the operating bandwidth becomes: 0.5 ≤ a/λ ≤ 1, or, a ≤ λ ≤ 2a.
Waveguide modes can be excited by inserting small probes at the beginning of the We will see later that the total amount of transmitted power in this mode is propor-
waveguide. The probes are chosen to generate an electric field that resembles the field tional to the cross-sectional area of the guide, ab. Thus, if in addition to having the
of the desired mode. † Murphy’s law for waveguides states that “if a mode can propagate, it will.”
8.7. Power Transfer, Energy Density, and Group Velocity 257 258 Electromagnetic Waves & Antennas – S. J. Orfanidis

where

jβ ω
H1 = H0 , E0 = −ηTE H1 = −jη H0 (8.7.2)
kc ωc
The Poynting vector is obtained from the general result of Eq. (8.3.7):

1 1 1
Pz = |ET |2 = |Ey (x)|2 = |E0 |2 sin2 kc x
2ηTE 2ηTE 2ηTE

The transmitted power is obtained by integrating Pz over the cross-sectional area

Fig. 8.6.1 Operating bandwidth in rectangular waveguides.
of the guide:

a b
widest bandwidth, we also require to have the maximum power transmitted, the dimen- 1
PT = |E0 |2 sin2 kc x dxdy
sion b must be chosen to be as large as possible, that is, b = a/2. Most practical guides 0 0 2ηTE
follow these side proportions.
Noting the definite integral,
If there is a “canonical” guide, it will have b = a/2 and be operated at a frequency
that lies in the middle of the operating band [fc , 2fc ], that is, a a  πx a
sin2 kc x dx = sin2 dx = (8.7.3)
c 0 0 a 2
f = 1.5fc = 0.75 (8.6.1) 
a and using ηTE = ηω/βc = η/ 1 − ω2c /ω2 , we obtain:
Table 8.6.1 lists some standard air-filled rectangular waveguides with their naming
designations, inner side dimensions a, b in inches, cutoff frequencies in GHz, minimum

and maximum recommended operating frequencies in GHz, power ratings, and attenua- 1 1 ω2c
2
tions in dB/m (the power ratings and attenuations are representative over each operating PT = |E0 | ab = |E0 |2 ab 1− (transmitted power) (8.7.4)
4ηTE 4η ω2
band.) We have chosen one example from each microwave band.
We may also calculate the distribution of electromagnetic energy along the guide, as
name a b fc fmin fmax band P α measured by the time-averaged energy density. The energy densities of the electric and
WR-510 5.10 2.55 1.16 1.45 2.20 L 9 MW 0.007 magnetic fields are:
WR-284 2.84 1.34 2.08 2.60 3.95 S 2.7 MW 0.019
WR-159 1.59 0.795 3.71 4.64 7.05 C 0.9 MW 0.043 1 1 1
WR-90 0.90 0.40 6.56 8.20 12.50 X 250 kW 0.110
we = Re E · E∗ = |Ey |2
2 2 4
WR-62 0.622 0.311 9.49 11.90 18.00 Ku 140 kW 0.176
1 1 1 
WR-42 0.42 0.17 14.05 17.60 26.70 K 50 kW 0.370 wm = Re µH · H∗ = µ |Hx |2 + |Hz |2
WR-28 0.28 0.14 21.08 26.40 40.00 Ka 27 kW 0.583 2 2 4
WR-15 0.148 0.074 39.87 49.80 75.80 V 7.5 kW 1.52
WR-10 0.10 0.05 59.01 73.80 112.00 W 3.5 kW 2.74 Inserting the expressions for the fields, we find:

1 1 
Table 8.6.1 Characteristics of some standard air-filled rectangular waveguides. we = |E0 |2 sin2 kc x , wm = µ |H1 |2 sin2 kc x + |H0 |2 cos2 kc x
4 4

Because these quantities represent the energy per unit volume, if we integrate them
over the cross-sectional area of the guide, we will obtain the energy distributions per
8.7 Power Transfer, Energy Density, and Group Velocity unit z-length. Using the integral (8.7.3) and an identical one for the cosine case, we find:

Next, we calculate the time-averaged power transmitted in the TE10 mode. We also calcu- a b a b
1 1
late the energy density of the fields and determine the velocity by which electromagnetic We = We (x, y) dxdy = |E0 |2 sin2 kc x dxdy = |E0 |2 ab
0 0 0 0 4 8
energy flows down the guide and show that it is equal to the group velocity. We recall
that the non-zero field components are:
a b
1  1 
Wm = µ |H1 |2 sin2 kc x + |H0 |2 cos2 kc x dxdy = µ |H1 |2 + |H0 |2 ab
0 0 4 8
Hz (x)= H0 cos kc x , Hx (x)= H1 sin kc x , Ey (x)= E0 sin kc x (8.7.1)
8.8. Power Attenuation 259 260 Electromagnetic Waves & Antennas – S. J. Orfanidis

Although these expressions look different, they are actually equal, We = Wm . In- The field expressions (8.4.3) were derived assuming the boundary conditions for
deed, using the property β2 /kc2 + 1 = (β2 + k2c )/k2c = k2 /k2c = ω2 /ω2c and the relation- perfectly conducting wall surfaces. The induced surface currents on the inner walls of
ships between the constants in (8.7.1), we find: the waveguide are given by Js = n̂ × H, where the unit vector n̂ is ±x̂ and ±ŷ on the
left/right and bottom/top walls, respectively.
  β2 ω2 µ
µ |H1 |2 + |H0 |2 = µ |H0 |2 2 + |H0 |2 = µ|H0 |2 2 = 2 |E0 |2 = |E0 |2 The surface currents and tangential magnetic fields are shown in Fig. 8.8.1. In par-
kc ωc η ticular, on the bottom and top walls, we have:
The equality of the electric and magnetic energies is a general property of wavegui-
ding systems. We also encountered it in Sec. 2.3 for uniform plane waves. The total
energy density per unit length will be:

1
W = We + Wm = 2We = |E0 |2 ab (8.7.5)
4
According to the general relationship between flux, density, and transport velocity
given in Eq. (1.5.2), the energy transport velocity will be the ratio ven = PT /W . Using
√
Eqs. (8.7.4) and (8.7.5) and noting that 1/η = 1/ µ = c, we find:

 Fig. 8.8.1 Currents on waveguide walls.

PT ω2
ven = =c 1 − c2 (energy transport velocity) (8.7.6)
W ω
Js = ±ŷ × H = ±ŷ × (x̂ Hx + ẑHz )= ±(−ẑ Hx + x̂ Hz )= ±(−ẑ H1 sin kc x + x̂ H0 cos kc x)
This is equal to the group velocity of the propagating mode. For any dispersion
relationship between ω and β, the group and phase velocities are defined by Similarly, on the left and right walls:

dω ω Js = ±x̂ × H = ±x̂ × (x̂ Hx + ẑHz )= ∓ŷ Hz = ∓ŷ H0 cos kc x

vgr = , vph = (group and phase velocities) (8.7.7)
dβ β
At x = 0 and x = a, this gives Js = ∓ŷ(±H0 )= ŷ H0 . Thus, the magnitudes of the
For uniform plane waves and TEM transmission lines, we have ω = βc, so that vgr = surface currents are on the four walls:
vph = c. For a rectangular waveguide, we have ω2 = ωc2 + β2 c2 . Taking differentials of 
|H0 |2 , (left and right walls)
both sides, we find 2ωdω = 2c2 βdβ, which gives: 2
|Js | =
|H0 |2 cos2 kc x + |H1 |2 sin2 kc x , (top and bottom walls)

dω βc2 ω2c The power loss per unit z-length is obtained from Eq. (8.2.8) by integrating |Js |2
vgr = = =c 1− (8.7.8)
dβ ω ω2 around the four walls, that is,

where we used Eq. (8.1.10). Thus, the energy transport velocity is equal to the group 1 a
1 b
velocity, ven = vgr . We note that vgr = βc2 /ω = c2 /vph , or Ploss = 2 Rs |Js |2 dx + 2 Rs |Js |2 dy
2 0 2 0

a b
vgr vph = c2 (8.7.9) = Rs |H0 |2 cos2 kc x + |H1 |2 sin2 kc x dx + Rs |H0 |2 dy
0 0
The energy or group velocity satisfies vgr ≤ c, whereas vph ≥ c. Information trans-
a Rs a  2b
mission down the guide is by the group velocity and, consistent with the theory of = Rs |H0 |2 + |H1 |2 + Rs b|H0 |2 = |H0 |2 + |H1 |2 + |H0 |2
relativity, it is less than c. 2 2 a

Using |H0 |2 +|H1 |2 = |E0 |2 /η2 from Sec. 8.7, and |H0 |2 = (|E0 |2 /η2 )ω2c /ω2 , which
follows from Eq. (8.4.2), we obtain:
8.8 Power Attenuation
Rs a|E0 |2 2b ω2c
In this section, we calculate the attenuation coefficient due to the ohmic losses of the Ploss = 1+
2η2 a ω2
conducting walls following the procedure outlined in Sec. 8.2. The losses due to the
filling dielectric can be determined from Eq. (8.2.5). The attenuation constant is computed from Eqs. (8.2.9) and (8.7.4):
8.8. Power Attenuation 261 262 Electromagnetic Waves & Antennas – S. J. Orfanidis

The cutoff frequency of the TE10 mode is fc = c/2a = 3.71 GHz. The maximum operating
bandwidth is the interval [fc , 2fc ]= [3.71, 7.42] GHz, and the recommended interval is
Rs a|E0 |2 2b ω2c [4.64, 7.05] GHz.
1+
Ploss 2η2 a ω2
αc = =  Assuming copper walls with conductivity σ = 5.8×107 S/m, the calculated attenuation
2PT 1 ωc 2 constant αc from Eq. (8.8.1) is plotted in dB/m versus frequency in Fig. 8.8.2.
2 |E0 |2 ab 1−
4η ω2
Attenuation Coefficient Power Transmitted
0.1 1.5
which gives:
bandwidth
0.08
2b ω2c
1+ 1
Rs a ω2


α (dB/m)

PT (MW)
αc = (attenuation of TE10 mode) (8.8.1) 0.06
ηb ω2c
1−
ω2 0.04
0.5

This is in units of nepers/m. Its value in dB/m is obtained by αdB = 8.686αc . For a 0.02
bandwidth

given ratio a/b, αc increases with decreasing b, thus the smaller the guide dimensions,
the larger the attenuation. This trend is noted in Table 8.6.1. 0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
The main tradeoffs in a waveguiding system are that as the operating frequency f f (GHz) f (GHz)
increases, the dimensions of the guide must decrease in order to maintain the operat-
ing band fc ≤ f ≤ 2fc , but then the attenuation increases and the transmitted power Fig. 8.8.2 Attenuation constant and transmitted power in a WR-159 waveguide.
decreases as it is proportional to the guide’s area.
The power transmitted PT is calculated from Eq. (8.7.4) assuming a maximum breakdown
Example 8.8.1: Design a rectangular air-filled waveguide to be operated at 5 GHz, then, re- voltage of E0 = 1.5 MV/m, which gives a safety factor of two over the dielectric breakdown
design it to be operated at 10 GHz. The operating frequency must lie in the middle of the of air of 3 MV/m. The power in megawatt scales is plotted in Fig. 8.8.2.
operating band. Calculate the guide dimensions, the attenuation constant in dB/m, and 
the maximum transmitted power assuming the maximum electric field is one-half of the Because of the factor 1 − ω2c /ω2 in the denominator of αc and the numerator of PT ,
dielectric strength of air. Assume copper walls with conductivity σ = 5.8×107 S/m. the attenuation constant becomes very large near the cutoff frequency, while the power is
almost zero. A physical explanation of this behavior is given in the next section. 
Solution: If f is in the middle of the operating band, fc ≤ f ≤ 2fc , where fc = c/2a, then
f = 1.5fc = 0.75c/a. Solving for a, we find
8.9 Reflection Model of Waveguide Propagation
0.75c 0.75×30 GHz cm
a= = = 4.5 cm
f 5
An intuitive model for the TE10 mode can be derived by considering a TE-polarized
For maximum power transfer, we require b = a/2 = 2.25 cm. Because ω = 1.5ωc , we
uniform plane wave propagating in the z-direction by obliquely bouncing back and forth
have ωc /ω = 2/3. Then, Eq. (8.8.1) gives αc = 0.037 dB/m. The dielectric strength of air between the left and right walls of the waveguide, as shown in Fig. 8.9.1.
is 3 MV/m. Thus, the maximum allowed electric field in the guide is E0 = 1.5 MV/m. Then, If θ is the angle of incidence, then the incident and reflected (from the right wall)
Eq. (8.7.4) gives PT = 1.12 MW. wavevectors will be:
At 10 GHz, because f is doubled, the guide dimensions are halved, a = 2.25 and b = 1.125 k = x̂ kx + ẑ kz = x̂ k cos θ + ẑ k sin θ
√
cm. Because Rs depends on f like f 1/2 , it will increase by a factor of 2. Then, the factor
√
Rs /b will increase by a factor of 2 2. Thus, the attenuation will increase to the value
√
k = −x̂ kx + ẑ kz = −x̂ k cos θ + ẑ k sin θ
αc = 0.037 · 2 2 = 0.104 dB/m. Because the area ab is reduced by a factor of four, so
will the power, PT = 1.12/4 = 0.28 MW = 280 kW. The electric and magnetic fields will be the sum of an incident and a reflected com-
The results of these two cases are consistent with the values quoted in Table 8.6.1 for the ponent of the form:
C-band and X-band waveguides, WR-159 and WR-90. 
E = ŷ E1 e−jk·r + ŷ E1 e−jk ·r = ŷ E1 e−jkx x e−jkz z + ŷ E1 ejkx x e−jkz z = E1 + E1
Example 8.8.2: WR-159 Waveguide. Consider the C-band WR-159 air-filled waveguide whose
1 1
characteristics were listed in Table 8.6.1. Its inner dimensions are a = 1.59 and b = a/2 = H= k̂ × E1 + k̂ × E1
0.795 inches, or, equivalently, a = 4.0386 and b = 2.0193 cm.
η η
8.9. Reflection Model of Waveguide Propagation 263 264 Electromagnetic Waves & Antennas – S. J. Orfanidis

The boundary condition on the right wall requires sin kx a = 0, which gives rise to
the same condition as (8.4.4), that is, kc a = nπ.
This model clarifies also the meaning of the group velocity. The plane wave is bounc-
ing left and right with the speed of light c. However, the component of this velocity in
the z-direction will be vz = c sin θ. This is equal to the group velocity. Indeed, it follows
from Eq. (8.9.3) that:

ω2c
vz = c sin θ = c 1− = vgr (8.9.5)
ω2
Eq. (8.9.3) implies also that at ω = ωc , we have sin θ = 0, or θ = 0, that is, the wave
is bouncing left and right at normal incidence, creating a standing wave, and does not
Fig. 8.9.1 Reflection model of TE10 mode. propagate towards the z-direction. Thus, the transmitted power is zero and this also
implies, through Eq. (8.2.9), that αc will be infinite.
On the other hand, for very large frequencies, ω  ωc , the angle θ will tend to 90o ,
where the electric field was taken to be polarized in the y direction. These field expres- causing the wave to zoom through guide almost at the speed of light.
sions become component-wise:

Ey = E1 e−jkx x + E1 ejkx x e−jkz z 8.10 Resonant Cavities
1 
Hx = − sin θ E1 e−jkx x + E1 ejkx x e−jkz z Cavity resonators are metallic enclosures that can trap electromagnetic fields. The
η (8.9.1)
boundary conditions on the cavity walls force the fields to exist only at certain quantized
1  resonant frequencies. For highly conducting walls, the resonances are extremely sharp,
Hz = cos θ E1 e−jkx x − E1 ejkx x e−jkz z
η having a very high Q of the order of 10,000.
The boundary condition on the left wall, x = 0, requires that E1 + E1 = 0. We may write Because of their high Q , cavities can be used not only to efficiently store electro-
therefore, E1 = −E1 = jE0 /2. Then, the above expressions simplify into: magnetic energy at microwave frequencies, but also to act as precise oscillators and to
perform precise frequency measurements.
Ey = E0 sin kx x e−jkz z Fig. 8.10.1 shows a rectangular cavity with z-length equal to l formed by replacing
the sending and receiving ends of a waveguide by metallic walls. A forward-moving wave
1
Hx = − sin θE0 sin kx x e−jkz z will bounce back and forth from these walls, resulting in a standing-wave pattern along
η (8.9.2)
the z-direction.
j
Hz = cos θE0 cos kx x e−jkz z
η
These are identical to Eq. (8.4.3) provided we identify β with kz and kc with kx , as
shown in Fig. 8.9.1. It follows from the wavevector triangle in the figure that the angle
of incidence θ will be given by cos θ = kx /k = kc /k, or,


ωc ω2c
cos θ = , sin θ = 1− (8.9.3)
ω ω2 Fig. 8.10.1 Rectangular cavity resonator (and induced wall currents for the TEn0p mode.)

The ratio of the transverse components, −Ey /Hx , is the transverse impedance, which
is recognized to be ηTE . Indeed, we have: Because the tangential components of the electric field must vanish at the end-walls,
these walls must coincide with zero crossings of the standing wave, or put differently, an
Ey η η integral multiple of half-wavelengths must fit along the z-direction, that is, l = pλg /2 =
ηTE = − = =  (8.9.4)
Hx sin θ ω2 pπ/β, or β = pπ/l, where p is a non-zero integer. For the same reason, the standing-
1 − c2
ω wave patterns along the transverse directions require a = nλx /2 and b = mλy /2, or
8.10. Resonant Cavities 265 266 Electromagnetic Waves & Antennas – S. J. Orfanidis

kx = nπ/a and ky = mπ/b. Thus, all three cartesian componentsof the wave vector The ohmic losses are calculated from Eq. (8.2.6), integrated over all six cavity sides.
are quantized, and therefore, so is the frequency of the wave ω = c k2x + k2y + β2 : The surface currents induced on the walls are related to the tangential magnetic fields
by J s = n̂ × Htan . The directions of these currents are shown in Fig. 8.10.1. Specifically,
 2  2  2 we find for the currents on the six sides:
nπ mπ pπ
ωnmp = c + + (resonant frequencies) (8.10.1) ⎧ 2
a b l ⎪ 2
⎨H0 sin βz
⎪ (left & right)
2
Such modes are designated as TEnmp or TMnmp . For simplicity, we consider the case |J s | = H0 cos2 kc x sin2 βz + H12 sin2 kc x cos2 βz
2
(top & bottom)
⎪
⎪
TEn0p . Eqs. (8.3.6) also describe backward-moving waves if one replaces β by −β, which
⎩ 2
H1 sin2 kc x (front & back)
also changes the sign of ηTE = ηω/βc. Starting with a linear combination of forward
and backward waves in the TEn0 mode, we obtain the field components: The power loss can be computed by integrating the loss per unit conductor area,
 Eq. (8.2.6), over the six wall sides, or doubling the answer for the left, top, and front
Hz (x, z) = H0 cos kc x Ae−jβz + Bejβz , sides. Using the integrals (8.10.5), we find:
 β  
Hx (x, z) = jH1 sin kc x Ae−jβz − Bejβz , H1 = H0 1 bl al ab
kc (8.10.2) Ploss = Rs |J s |2 dA = Rs H02 + (H02 + H12 ) + H12
2 walls 2 4 2
 ω   (8.10.6)
Ey (x, z) = −jE0 sin kc x Ae−jβz + Bejβz , E0 = ηH0 1 β 2
ωc = Rs H02 l(2b + a)+ 2 a(2b + l)
4 kc
where ωc = ckc . By requiring that Ey (x, z) have z-dependence of the form sin βz, the
coefficients A, B must be chosen as A = −B = j/2. Then, Eq. (8.10.2) specializes into: where we substituted H12 = H02 β2 /k2c . It follows that the Q -factor will be:
Hz (x, z) = H0 cos kc x sin βz ,
W ωµ (k2c + β2 )(abl)
Q=ω =
β Ploss 2Rs k2c l(2b + a)+β2 a(2b + l)
Hx (x, z) = −H1 sin kc x cos βz , H1 = H0 (8.10.3)
kc
For the TEn0p mode we have β = pπ/l and kc = nπ/a. Using Eq. (8.2.7) to replace
ω
Ey (x, z) = −jE0 sin kc x sin βz , E0 = ηH0 Rs in terms of the skin depth δ, we find:
ωc
As expected, the vanishing of Ey (x, z) on the front/back walls, z = 0 and z = l, and n2 p2
1
+
on the left/right walls, x = 0 and x = a, requires the quantization conditions: β = pπ/l Q= a2 l2
    (8.10.7)
and kc = nπ/a. The Q of the resonator can be calculated from its definition: δ n2 2 1 p2 2 1
+ + +
a2 a b l2 l b
W
Q=ω (8.10.4)
Ploss The lowest resonant frequency corresponds to n = p = 1. For a cubic cavity, a =
b = l, the Q and the lowest resonant frequency are:
where W is the total time-averaged energy stored within the cavity volume and Ploss is
the total power loss due to the wall ohmic losses (plus other losses, such as dielectric
√
a cπ 2 ω c
losses, if present.) The ratio ∆ω = Ploss /W is usually identified as the 3-dB width of the Q= , ω101 = , f101 = = √ (8.10.8)
3δ a 2π a 2
resonance centered at frequency ω. Therefore, we may write Q = ω/∆ω.
It is easily verified that the electric and magnetic energies are equal, therefore, W For an air-filled cubic cavity with a = 3 cm, we find f101 = 7.07 GHz, δ = 7.86×10−5
may be calculated by integrating the electric energy density over the cavity volume: cm, and Q = 12724. As in waveguides, cavities can be excited by inserting small probes
a b l
that generate fields resembling a particular mode.
1 1
W = 2We = 2 |Ey (x, z)|2 dx dy dz = |E0 |2 sin2 kc x cos2 βz dx dy dz
4 vol 2 0 0 0
 
1 1 ω2 1 k2c + β2 8.11 Dielectric Slab Waveguides
= |E0 |2 (abl)= µ|H0 |2 2 (abl)= µ |H0 |2 (abl)
8 8 ωc 8 k2c
A dielectric slab waveguide is a planar dielectric sheet or thin film of some thickness,
where we used the following definite integrals (valid because kc = nπ/a, β = pπ/l) : say 2a, as shown in Fig. 8.11.1. Wave propagation in the z-direction is by total internal
a a l l reflection from the left and right walls of the slab. Such waveguides provide simple
a l
sin2 kc x dx = cos2 kc x dx = , sin2 βz dz = cos2 βz dz = (8.10.5) models for the confining mechanism of waves propagating in optical fibers.
0 0 2 0 0 2
8.11. Dielectric Slab Waveguides 267 268 Electromagnetic Waves & Antennas – S. J. Orfanidis

k2c = k20 n21 − β2 k2c = k20 n21 − β2

⇒ (8.11.3)
−α2c = k20 n22 − β2 α2c = β2 − k20 n22

Similarly, Eqs. (8.11.2) read:

∂2x Hz (x)+k2c Hz (x)= 0 for |x| ≤ a

(8.11.4)
∂2x Hz (x)−α2c Hz (x)= 0 for |x| ≥ a
The two solutions sin kc x and cos kc x inside the guide give rise to the so-called even
and odd TE modes (referring to the even-ness or oddness of the resulting electric field.)
For the even modes, the solutions of Eqs. (8.11.4) have the form:
Fig. 8.11.1 Dielectric slab waveguide.
⎧
⎪
⎪ H sin kc x , if −a ≤ x ≤ a
⎨ 1
The propagating fields are confined primarily inside the slab, however, they also Hz (x)= H2 e−αc x , if x≥a (8.11.5)
⎪
⎪
exist as evanescent waves outside it, decaying exponentially with distance from the slab. ⎩
H3 eαc x , if x ≤ −a
Fig. 8.11.1 shows a typical electric field pattern as a function of x.
For simplicity, we assume that the media to the left and right of the slab are the The corresponding x-components are obtained by applying Eq. (8.3.8) using the ap-
same. To guarantee total internal reflection, the dielectric constants inside and outside propriate value for k2c , that is, k2c2 = −α2c outside and k2c1 = k2c inside:
the slab must satisfy 1 > 2 , and similarly for the refractive indices, n1 > n2 . ⎧
⎪
⎪ jβ jβ
We look for TE solutions that depend only on the x coordinate. The cutoff wavenum- ⎪
⎪ − 2 ∂x Hz (x)= − H1 cos kc x , if −a ≤ x ≤ a
⎪
⎪ k kc
⎪
⎪ c
ber kc appearing in the Helmholtz equation for Hz (x) depends on the dielectric constant ⎨ jβ jβ
of the propagation medium, k2c = ω2 µ−β2 . Therefore, k2c takes different values inside Hx (x)= − 2 ∂x Hz (x)= − H2 e−αc x , if x≥a (8.11.6)
⎪
⎪ −α α
⎪
⎪ c c
and outside the guide: ⎪
⎪
⎪
⎪ jβ jβ
⎩ − ∂x Hz (x)= H3 eαc x , if x≥a
−α2c αc
k2c1 = ω2 1 µ0 − β2 = ω2 0 µ0 n21 − β2 = k20 n21 − β2 (inside)
(8.11.1) The electric fields are Ey (x)= −ηTE Hx (x), where ηTE = ωµ0 /β is the same inside
k2c2 = ω 2 µ0 − β = ω
2 2 2
0 µ0 n22 −β =
2
k20 n22 −β
2
(outside) and outside the slab. Thus, the electric field has the form:

where k0 = ω/c0 is the free-space wavenumber. We note that ω, β are the same inside ⎧
⎪
⎪ E cos kc x , if −a ≤ x ≤ a
and outside the guide. This follows from matching the tangential fields at all times t ⎨ 1
Ey (x)= E2 e−αc x , if x≥a (even TE modes) (8.11.7)
and all points z along the slab walls. The corresponding Helmholtz equations in the ⎪
⎪
⎩
regions inside and outside the guide are: E3 eαc x , if x ≤ −a
where we defined the constants:
∂2x Hz (x)+k2c1 Hz (x)= 0 for |x| ≤ a
(8.11.2)
jβ jβ jβ
∂2x Hz (x)+k2c2 Hz (x)= 0 for |x| ≥ a E1 = ηTE H1 , E2 = ηTE H2 , E3 = − ηTE H3 (8.11.8)
kc αc αc
Inside the slab, the solutions are sin kc1 x and cos kc1 x, and outside, sin kc2 x and
The boundary conditions state that the tangential components of the magnetic and
cos kc2 x, or equivalently, e±jkc2 x . In order for the waves to remain confined in the near
electric fields, that is, Hz , Hx , Ey , are continuous across the dielectric interfaces at x =
vicinity of the slab, the quantity kc2 must be imaginary, for if it is real, the fields would
−a and x = a. Because Ey = −ηTE Hx and ηTE is the same in both media, the continuity
propagate at large x distances from the slab (they would correspond to the rays refracted
of Ey follows from the continuity of Hx . The continuity of Hz at x = a and x = −a
from the inside into the outside.)
implies that:
If we set kc2 = −jαc , the solutions outside will be e±αc x . If αc is positive, then only
the solution e−αc x is physically acceptable to the right of the slab, x ≥ a, and only eαc x
H1 sin kc a = H2 e−αc a and − H1 sin kc a = H3 e−αc a (8.11.9)
to the left, x ≤ −a. Thus, the fields attenuate exponentially with the transverse distance
x, and exist effectively within a skin depth distance 1/αc from the slab. Setting kc1 = kc Similarly, the continuity of Hx implies (after canceling a factor of −jβ):
and kc2 = −jαc , Eqs. (8.11.1) become in this new notation:
8.11. Dielectric Slab Waveguides 269 270 Electromagnetic Waves & Antennas – S. J. Orfanidis

⎧
1 1 1 1 ⎪
⎪ E cos kc x , if −a ≤ x ≤ a
H1 cos kc a = H2 e−αc a and H1 cos kc a = − H3 e−αc a (8.11.10) ⎨ 1
kc αc kc αc E1 cos kc a e−αc (x−a) , if x≥a
Ey (x)= (even TE modes) (8.11.19)
⎪
⎪
⎩
Eqs. (8.11.9) and (8.11.10) imply: E1 cos kc a eαc (x+a) , if x ≤ −a

αc
H2 = −H3 = H1 eαc a sin kc a = H1 eαc a cos kc a (8.11.11) ⎧
kc ⎪
⎪ E sin kc x , if −a ≤ x ≤ a
⎨ 1
Ey (x)= E1 sin kc a e−αc (x−a) , if x≥a (odd TE modes) (8.11.20)
Similarly, we find for the electric field constants: ⎪
⎪
⎩
−E1 sin kc a eαc (x+a) , if x ≤ −a
kc
E2 = E3 = E1 eαc a cos kc a = E1 eαc a sin kc a (8.11.12) Given the operating frequency ω, Eqs. (8.11.3) and (8.11.13) or (8.11.18) provide three
αc
equations in the three unknowns kc , αc , β. To solve them, we add the two equations
The consistency of the last equations in (8.11.11) or (8.11.12) requires that: (8.11.3) to eliminate β:

kc
cos kc a = sin kc a ⇒ αc = kc tan kc a (8.11.13) ω2 2
αc α2c + k2c = k20 (n21 − n22 )= (n1 − n22 ) (8.11.21)
c20
For the odd TE modes, we have for the solutions of Eq. (8.11.4):
⎧ Next, we discuss the numerical solutions of these equations. Defining the dimen-
⎪
⎪ H cos kc x , if −a ≤ x ≤ a sionless quantities u = kc a and v = αc a, we may rewrite Eqs. (8.11.13), (8.11.18), and
⎨ 1
Hz (x)= H2 e−αc x , if x≥a (8.11.14) (8.11.21) in the equivalent forms:
⎪
⎪
⎩
H3 eαc x , if x ≤ −a
The resulting electric field is: v = u tan u v = −u cot u
(even modes) , (odd modes) (8.11.22)
⎧ v +u =R
2 2 2
v2 + u2 = R2
⎪
⎪ E sin kc x , if −a ≤ x ≤ a
⎨ 1
Ey (x)= E2 e−αc x , if x≥a (odd TE modes) (8.11.15) where R is the normalized frequency variable:
⎪
⎪
⎩
E3 eαc x , if x ≤ −a
ωa 2πf a 2πa
The boundary conditions imply in this case: R = k0 aNA = NA = NA = NA (8.11.23)
c0 c0 λ
αc 
H2 = H3 = H1 eαc a cos kc a = −H1 eαc a sin kc a (8.11.16) where NA = n21 − n22 is the numerical aperture of the slab and λ = c0 /f , the free-space
kc
wavelength.
and, for the electric field constants: Because the functions tan u and cot u have many branches, there may be several
possible solution pairs u, v for each value of R. These solutions are obtained at the
kc intersections of the curves v = u tan u and v = −u cot u with the circle of radius R,
E2 = −E3 = E1 eαc a sin kc a = −E1 eαc a cos kc a (8.11.17)
αc that is, v2 + u2 = R2 . Fig. 8.11.2 shows the solutions for various values of the radius R
corresponding to various values of ω.
The consistency of the last equation requires:
It is evident from the figure that for small enough R, that is, 0 ≤ R < π/2, there
is only one solution and it is even (for an optical fiber, the single-mode condition reads
αc = −kc cot kc a (8.11.18)
2πaNA /λ < 2.405, where a is the core radius.) For π/2 ≤ R < π, there are two
We note that the electric fields Ey (x) given by Eqs. (8.11.7) and (8.11.15) are even or solutions, one even and one odd. For π ≤ R < 3π/2, there are three solutions, two
odd functions of x for the two families of modes. Expressing E2 and E3 in terms of E1 , even and one odd, and so on. In general, there will be M + 1 solutions, alternating
we summarize the forms of the electric fields in the two cases: between even and odd, if R falls in the interval:

Mπ (M + 1)π
≤R< (8.11.24)
2 2
8.11. Dielectric Slab Waveguides 271 272 Electromagnetic Waves & Antennas – S. J. Orfanidis

v = u tan(u − Rm )
(mth mode) (8.11.29)
v2 + u2 = R2

If one had an approximate solution u, v for the mth mode, one could refine it by using
Newton’s method, which converges very fast provided it is close to the true solution. Just
such an approximate solution, accurate to within one percent of the true solution, was
given by Lotspeich [444]. Without going into the detailed justification of this method,
the approximation is as follows:

u = Rm + w1 (m)u1 (m)+w2 (m)u2 (m) , m = 0, 1, . . . , M (8.11.30)

where u1 (m), u2 (m) are approximate solutions near and far from the cutoff Rm , and
w1 (m), w2 (m) are weighting factors:

Fig. 8.11.2 Even and odd TE modes at different frequencies.
1 + 2R(R − Rm ) − 1 π R−m
u1 (m)= , u2 (m)=
R 2 R+1

w1 (m)= exp −(R − Rm )2 /Vm2
, w2 (m)= 1 − w1 (m) (8.11.31)
Given a value of R, we determine M as that integer satisfying Eq. (8.11.24), or, M ≤
 
2R/π < M + 1, that is, the largest integer less than 2R/π: 1 π/4 + Rm
Vm = √ − Rm
ln 1.25 cos(π/4)
 
2R This solution serves as the starting point to Newton’s iteration for solving the equa-
M = floor (maximum mode number) (8.11.25)
π tion F(u)= 0, where F(u) is defined by
Then, there will be M+ 1 solutions indexed by m = 0, 1, . . . , M, which will correspond 
to even modes if m is even and to odd modes if m is odd. The M + 1 branches of tan u F(u)= u tan(u − Rm )−v = u tan(u − Rm )− R2 − u2 (8.11.32)
and cot u being intersected by the R-circle are those contained in the u-ranges: Newton’s iteration is:

Rm ≤ u < Rm+1 , m = 0, 1, . . . , M (8.11.26)

for i = 1, 2 . . . , Nit do:
where F(u) (8.11.33)
u=u−
G(u)
mπ
Rm = , m = 0, 1, . . . , M (8.11.27)
where G(u) is the derivative F (u), correct to order O(F):
2
If m is even, the u-range (8.11.26) defines a branch of tan u, and if m is odd, a branch v u R2
of cot u. We can combine the even and odd cases of Eq. (8.11.22) into a single case by G(u)= + + (8.11.34)
u v u
noting the identity:
The solution steps defined in Eqs. (8.11.29)–(8.11.34) have been implemented in the
⎧ MATLAB function dslab.m, with usage:
⎨ tan u , if m is even
tan(u − Rm )= (8.11.28)
⎩ − cot u , if m is odd [u,v,err] = dslab(R,Nit); % TE-mode cutoff wavenumbers in a dielectric slab

This follows from the trigonometric identity: where Nit is the desired number of Newton iterations (8.11.33), err is the value of F(u)
at the end of the iterations, and u, v are the (M + 1)-dimensional vectors of solutions.
sin u cos(mπ/2)− cos u sin(mπ/2)
tan(u − mπ/2)= The number of iterations is typically very small, Nit = 2–3.
cos u cos(mπ/2)+ sin u sin(mπ/2)
The related MATLAB function dguide.m uses dslab to calculate the solution param-
Therefore, to find the mth mode, whether even or odd, we must find the unique eters β, kc , αc , given the frequency f , the half-length a, and the refractive indices n1 , n2
solution of the following system in the u-range Rm ≤ u < Rm+1 : of the slab. It has usage:
8.11. Dielectric Slab Waveguides 273 274 Electromagnetic Waves & Antennas – S. J. Orfanidis

[be,kc,ac,fc,err] = dguide(f,a,n1,n2,Nit); % dielectric slab guide outside the slab to be less confined. The electric field patterns are also shown in the figure
as functions of x.
where f is in GHz, a in cm, and β, kc , αc in cm−1 . The quantity fc is the vector of The approximation error, err, is found to be 4.885×10−15 using only three Newton itera-
the M + 1 cutoff frequencies defined by the branch edges Rm = mπ/2, that is, Rm = tions. Using two, one, and no (the Lotspeich approximation) iterations would result in the
ωm aNA /c0 = 2πfm aNA /c0 = mπ/2, or, errors 2.381×10−8 , 4.029×10−4 , and 0.058.
The lowest non-zero cutoff frequency is f1 = 8.6603 GHz, implying that there will be a
mc0
fm = , m = 0, 1, . . . , M (8.11.35) single solution if f is in the interval 0 ≤ f < f1 . For example, if f = 5 GHz, the solution is
4aNA
β = 1.5649 rad/cm, kc = 1.3920 rad/cm, and αc = 1.1629 nepers/cm.
The meaning of fm is that there are m + 1 propagating modes for each f is in the The frequency range over which there are only four solutions is [25.9808, 34.6410] GHz,
interval fm ≤ f < fm+1 . where the upper limit is 4f1 . 

Example 8.11.1: Dielectric Slab Waveguide. Determine the propagating TE modes of a dielectric In terms of the ray picture of the propagating wave, the angles of total internal
slab of half-length a = 0.5 cm at frequency f = 30 GHz. The refractive indices of the slab reflection are quantized according to the values of the propagation wavenumber β for
and the surrounding dielectric are n1 = 2 and n2 = 1. the various modes.
Solution: The solution is obtained by the MATLAB call: If we denote by k1 = k0 n1 the wavenumber within the slab, then the wavenumbers
β, kc are the z- and x-components kz , kx of k1 with an angle of incidence θ. (The vectorial
f = 30; a = 0.5; n1 = 2; n2 = 1; Nit = 3; relationships are the same as those in Fig. 8.9.1.) Thus, we have:
[be,kc,ac,fc,err] = dguide(f,a,n1,n2,Nit)
β = k1 sin θ = k0 n1 sin θ
(8.11.36)
The frequency radius is R = 5.4414, which gives 2R/π = 3.4641, and therefore, M = 3. kc = k1 cos θ = k0 n1 cos θ
The resulting solutions, depicted in Fig. 8.11.3, are as follows:
The value of β for each mode will generate a corresponding value for θ. The at-
TE Modes for R = 5.44 Electric Fields tenuation wavenumber αc outside the slab can also be expressed in terms of the total
7 internal reflection angles:
6 0 1
 
1
0 αc = β2 − k20 n22 = k0 n21 sin2 θ − n22
5 1
2
Since the critical angle is sin θc = n2 /n1 , we may also express αc as:
Ey (x) / E1

4 2 3
0 
v

3
αc = k0 n1 sin2 θ − sin θ2c (8.11.37)
2 3

−1
Example 8.11.2: For the Example 8.11.1, we calculate k0 = 6.2832 and k1 = 12.5664 rad/cm.
1
The critical and total internal reflection angles of the four modes are found to be:
0
0 1 2 3 4 5 6 7 −3 −2 −1 0 1 2 3  
u x/a n2
θc = asin = 30o
n1
 
Fig. 8.11.3 TE modes and corresponding E-field patterns. β
θ = asin = {77.8275o , 65.1960o , 51.5100o , 36.0609o }
k1

m u v β kc αc fm As required, all θs are greater than θc . 

0 1.3248 5.2777 12.2838 2.6497 10.5553 0.0000
1 2.6359 4.7603 11.4071 5.2718 9.5207 8.6603
2 3.9105 3.7837 9.8359 7.8210 7.5675 17.3205 8.12 Problems
3 5.0793 1.9519 7.3971 10.1585 3.9037 25.9808
8.1 An air-filled 1.5 cm×3 cm waveguide is operated at a frequency that lies in the middle of its
TE10 mode band. Determine this operating frequency in GHz and calculate the maximum
The cutoff frequencies fm are in GHz. We note that as the mode number m increases, power in Watts that can be transmitted without causing dielectric breakdown of air. The
the quantity αc decreases and the effective skin depth 1/αc increases, causing the fields dielectric strength of air is 3 MV/m.
8.12. Problems 275

8.2 It is desired to design an air-filled rectangular waveguide such that (a) it operates only in the
TE10 mode with the widest possible bandwidth, (b) it can transmit the maximum possible
power, and (c) the operating frequency is 12 GHz and it lies in the middle of the operating
band. What are the dimensions of the guide in cm?
8.3 An air-filled rectangular waveguide is used to transfer power to a radar antenna. The guide
must meet the following specifications: The two lowest modes are TE10 and TE20 . The op-
erating frequency is 3 GHz and must lie exactly halfway between the cutoff frequencies of
these two modes. The maximum electric field within the guide may not exceed, by a safety
margin of 3, the breakdown field of air 3 MV/m.

a. Determine the smallest dimensions a, b for such a waveguide, if the transmitted power
is required to be 1 MW.
b. What are the dimensions a, b if the transmitted power is required to be maximum?
What is that maximum power in MW?

8.4 It is desired to design an air-filled rectangular waveguide operating at 5 GHz, whose group
velocity is 0.8c. What are the dimensions a, b of the guide (in cm) if it is also required to carry
maximum power and have the widest possible bandwidth? What is the cutoff frequency of
the guide in GHz and the operating bandwidth?
8.5 Show the following relationship between guide wavelength and group velocity in an arbitrary
air-filled waveguide: vg λg = cλ , where λg = 2π/β and λ is the free-space wavelength.
Moreover, show that the λ and λg are related to the cutoff wavelength λc by:

1 1 1
= +
λ2 λ2g λ2c

8.6 Determine the four lowest modes that can propagate in a WR-159 and a WR-90 waveguide.
Calculate the cutoff frequencies (in GHz) and cutoff wavelengths (in cm) of these modes.
8.7 An air-filled WR-90 waveguide is operated at 9 GHz. Calculate the maximum power that
can be transmitted without causing dielectric breakdown of air. Calculate the attenuation
constant in dB/m due to wall ohmic losses. Assume copper walls.
8.8 A rectangular waveguide has sides a, b such that b ≤ a/2. Determine the cutoff wavelength
λc of this guide. Show that the operating wavelength band of the lowest mode is 0.5λc ≤
√
λ ≤ λc . Moreover, show that the allowed range of the guide wavelength is λg ≥ λc / 3.
8.9 The TE10 mode operating bandwidth of an air-filled waveguide is required to be 4–7 GHz.
What are the dimensions of the guide?
8.10 Computer Experiment: WR-159 Waveguide. Reproduce the two graphs of Fig. 8.8.2.
8.11 Computer Experiment: Dielectric Slab Waveguide. Using the MATLAB functions dslab and
dguide, write a program that reproduces all the results and graphs of Examples 8.11.1 and
8.11.2.
8.12 A TM mode is propagated along a waveguide of arbitrary but uniform cross section. Assume
perfectly conducting walls.

a. Show that the Ez (x, y) component satisfies:

∇Ez |2 dS = k2c
|∇ |Ez |2 dS
S S

b. Using the above result, show that the energy velocity is equal to the group velocity.