New waveguide technology for

antennas and circuits

E. Alfonso, A. Valero, J.I. Herranz, M. Baquero, M. Ferrando, V. Rodrigo, F. Vico,
E. Antonino, M. Cabedo, D. Sánchez, B. Bernardo, A. Vila
(1) Instituto de Telecomunicaciones y Aplicaciones Multimedia,
Universitat Politècnica de València
8G Building - access D - Camino de Vera s/n - 46022 Valencia (Spain)
Corresponding author: avalero@dcom.upv.es

Abstract new topologies of guiding structures, which are

especially suitable in many respects for frequen-
This review paper describes a new type of met- cies over 30 GHz, stressing their properties and
amaterial-based quasi-TEM waveguiding device. giving a physical insight. Their application to an-
The transmission line is based on the principle tennas and circuits is also shown through prelimi-
that combining two parallel plates, one being a nary designs, some of which have already been
perfect electric conductor and the other a per- validated experimentally.
fect magnetic conductor, provides a stop band
area between them as long as the separation be- Our first incursion in the study and research of
tween plates is less than a quarter wavelength. In new waveguides for high frequencies is caused
that scenario, metallic inclusions such as strips, by the difficulties found in the manufacturing of
ridges or grooves open a propagation path in waveguide slot arrays for steerable antennas at
the gap between the inclusion and the metallic frequencies over 30 GHz. One of the most com-
conductor and at the same time fields remain mon ways of feeding slot arrays is by means of
confined without resorting to vertical walls. The a group of juxtaposed monomode rectangular
paper explains this concept in detail; it describes waveguides. This solution is generally adopted
its advantages and shows several application ex- for antennas requiring control of the beam steer-
amples and experimental demonstrations. ing. However, it presents serious problems in the
manufacturing process derived from the physi-
Keywords Gap waveguides, metamaterial- cal constraints imposed by the design itself to
based waveguides, hard surfaces, microwave meet antenna specifications. As it is well-known,
circuits, millimeter-wave antennas, slot arrays, from array theory, elements spacing must not
waveguide slot arrays. exceed a maximum distance to avoid grating
lobes. This maximum distance is even smaller
when the radiated beam is scanned. On the other
1. Introduction hand, rectangular waveguides cannot be as nar-
row as desired because they could be below cut-
An increasing demand of applications at the mil- off. Both requirements force the waveguides to
limeter- and submillimeter-wave bands is expect- work close to the cut-off frequency, where losses
ed to be imminent. There are already some ap- at sidewalls are higher, especially at the millim-
plications working at these frequencies, but they eter- and submillimeter-bands, and make also
suffer from the limitations of the existing technol- the internal walls to be very thin, what means
ogies. There is then a real need for new waveguide that a milling technique could be not the most
technologies which are more appropriate for high suitable one for the fabrication. Besides, these
frequency applications. This paper presents two waveguide slot arrays are usually made in two

Waves · 2011 · year 3 65

 Oversized rectangular waveguides can support
a multiplicity of modes, each one with its own
propagation constant. By a proper combination
of TE modes it is possible to generate a uniform
quasi-TEM mode [1]-[2], but its inherent disper-
sive behavior makes this quasi-TEM propagation
very narrowband. As it is well-known, each TE
mode in a waveguide can be seen as the result of
the interference between two waves impinging
and reflecting successively from both sidewalls
with a particular angle of incidence [3]. Thus, it
is easy to glimpse that if transverse currents are
stopped so that only longitudinal currents can
flow along the waveguide, then the two plane
Figure 1. Top view of an oversized rectangular waveguide excited by a single waves which form each of the TE modes are
probe. Magnitude of the transverse x-component of the magnetic field inside this forced to change their propagation directions
waveguide and also in case of transverse currents stopped. to the longitudinal direction of the waveguide
(see Fig. 1). Thereby, only plane waves parallel to
the longitudinal direction are allowed. In other
parts, the bottom part consisting of a groove
words, any kind of propagation is suppressed
surface, and the top part which is a metal plate
except for the TEM-like one.
containing the radiating slots. Both parts have to
be joined together later assuring a good electri-
The previous reasoning leads us to try to find out
cal contact at joints. The previous drawback add-
a waveguide in which propagation in the trans-
ed to the fact that the plate is never completely
verse direction is inhibited while it is favored in
flat, presenting protrusions and irregularities,
the longitudinal direction. As a result, the Single
lead to a difficult manufacturing process which
Hard Wall Waveguide (SHWW) comes up [4]-[6].
increases the final cost of making these anten-
The waveguide is comprised of a conducting
nas, only affordable for military applications.
upper face and a hard surface [7] as the lower
face. The gap between these two faces, d, must
Several attempts have been made at the Tokio In-
not be higher than a quarter wavelength. The
stitute of Technology to solve the junction prob-
waveguide can be as many wavelengths wide
lem. One proposal consists of alternating phase-
as desired and its left and right sides can be left
fed waveguides, where adjacent waveguides are
open since walls are not required to support the
fed with opposite phase. The net current at joints
modes. Nevertheless, conducting sidewalls may
is zero, therefore a perfect electric contact is not
be used, closing the waveguide to assure rigidity
required. Other solution was based on oversized
of the whole structure, as sketched in Fig. 2.
waveguides, electrically wide, fed by a quasi-
TEM wave. However, it is a narrow band solution
Single hard wall waveguides are a new type of
and not valid at all for phase-scanning.
waveguides that are able to propagate a plurali-
ty of degenerate independent quasi-TEM modes
For the purpose of devising a low-cost solu-
within the same waveguide. Each of these local
tion to feed slot arrays, the first idea was to
modes would be excited by its own source (see
use an oversized rectangular waveguide, i.e., a
Fig. 3). The waveguide does not require vertical
waveguide that is much wider than the standard
walls to isolate these parallel modes from each
monomode ones. But now, some kind of mate-
other since they experience a strong transverse
rial or periodic structure should be added to the
attenuation. This property makes the waveguide
waveguide in order to modify its propagation
very attractive for applications over 30 GHz. In
properties in such a way that this waveguide
particular, those applications requiring multiple
emulates the behavior of a collection of juxta-
rectangular waveguides, like planar slot array
posed monomode rectangular waveguides. This
antennas or slotted waveguide power amplifi-
new structure should be simpler from the manu-
ers, can benefit from this type of waveguiding
facturing point of view.
structure.

Figure 2. Sketch of a Single Hard Wall Waveguide Figure 3. Four local quasi-TEM waves excited by
(SHWW). four probes within a SHWW.

66 ISSN 1889-8297 / Waves · 2011 · year 3

This idea has been later improved so that the Artificially soft and hard surfaces in electromag-
waves are no longer constrained to follow netics are realized by corrugations or by metal There is a real
straight paths, as shown in Fig. 4. The wave fol- strips loading a grounded substrate [14] (see Fig.
need for new
lows in the gap between a conducting ridge 5). Such surfaces make use of surface resonances
and the metallic upper face because the ridge owing to the depths of the grooves, appearing for
waveguide
is surrounded by artificial PMC surfaces on both depths of , where is the effective permittivity of technologies
sides, e.g., in the form of a bed of nails [8], i.e., the dielectric material, which is for corrugated which
a surface formed by metal pins standing on a soft surfaces and for the hard surface. In the soft are more
ground plane. The bed of nails is more effective case, corrugations are oriented transversely to the appropriate
than the grooves when the ridge supporting propagation and they form so-called electric and for high
the wave has to split, bend, make a corner, etc. magnetic current fences that stop the waves. On frequency
Therefore, this other new type of waveguide, the contrary, in the hard case, corrugations form applications.
so-called Ridge Gap Waveguide (RGW) [9]-[11], electric and magnetic current lines that support
is especially suitable to make millimeter- and wave propagation. Actually, the corrugated sur-
submillimeter-wave circuits and components. face used in this waveguide must act simultane-
ously as a hard surface in the direction of propa-
gation and as a soft surface in the transverse
direction. Yet the name given to the waveguide,
single hard wall waveguide, put the emphasis on
type of surface in the propagation direction.

x
z

Figure 4. Bottom part of a circuit in Ridge Gap d

Waveguide (RGW). h

2. Single hard wall waveguides

Figure 5. Single hard wall waveguide made up
of dielectric-filled corrugations. A sketch of the quasi-
2.1. Propagation properties
TEM wave is also shown.
Propagation in hollow waveguides is gener-
ally described in terms of eigenmodes. As it
is well-known, the collection of eigenmodes From the description above, it might be drawn
that can propagate depends on the waveguide that the conducting plate on top plays a second-
cross-section and the working frequency. Let ary role, since all the guiding seems to be done
us refer to the eigenmodes as global waves by the corrugated surface. However, the truth
to emphasize the fact that these solutions of is that the height of the gap left between the
Helmholtz's wave equation extend to the whole plate and the hard surface decisively contributes
cross-section of the waveguide. Let us now in- to keep waves bounded despite the absence
troduce local waves. Local waves, unlike global of walls. Indeed the plate must be less than a
waves, would be field solutions that propagate quarter wavelength above the hard surface to
in a highly confined fashion, using mainly a achieve the desired effect. To better understand
certain portion of the waveguide cross-section. this effect we should look at the ideal model of a
Certainly these waves are the result of a linear hard surface. Ideally, a hard surface can be seen
combination of eigenmodes and extend also to as a grid of parallel strips where every second
the whole waveguide cross-section, but they are strip is a perfect electric conductor (PEC) and a
negligible for most part of it [12]. Consequently, perfect magnetic conductor (PMC), respectively
if we consider an electrically wide waveguide, [15]. It can be easily shown that by placing the
several of these local waves could share the metallic upper face not higher than a quarter
same waveguide using different portions of the wavelength from the hard surface, propagation
waveguide cross-section without resorting to in any direction other than the longitudinal one
walls to keep them separated [13]. Single hard is stopped. A PEC strip combined with the PEC
wall waveguides have such a particular property; plate above forms a transmission line. On the
thereby it is possible to propagate multiple local contrary, when a wave encounters a PEC on the
waves within the same waveguide, as shown in upper plate and a PMC strip on the lower, it can
Fig. 3. These waves may show some degree of barely trespass it for the given gap height. After
overlapping depending on the spacing between crossing a few of these PEC/PMC strips in the
them, but beyond a given separation they can be transverse direction, the wave is deeply attenu-
considered independent. To drive waves locally, ated, producing a confined wave propagating
the waveguide is formed in the gap between a only in the longitudinal direction as depicted in
conducting plate and a hard surface. Fig. 5.

Waves · 2011 · year 3 / ISSN 1889-8297 67

 direction around the frequency where the hard
500
condition holds. As a result, fields excited by a

Propagation constant (m−1 )

450
given local source are forced to propagate lo-
400
350
cally in the corrugations direction, which means
300
that inner walls are not required to isolate waves.
250 Thereby, the manufacturing process of planar
200 slot array antennas can be greatly simplified if
we can avoid the walls between the waveguides,
150 ine
tl which become increasingly thinner as we go
100 il gh
50 above 30 GHz.
0
0 5 10 15 20
Frequency (GHz) The presence of four local quasi-TEM waves shar-
ing the waveguide cross-section is shown in Fig.
Figure 6. Dispersion diagram in the hard direc- 7. These waves are excited by four source probes
tion of a SHWW with a groove-filling material of fed with a uniform magnitude and a linear phase
εr=4. Groove depth is h = 2.95 mm, edge width s = of 45º. The probes are equally spaced apart 0.7λ, a
1 mm, corrugation period p = 3 mm. The gap be- typical spacing between rows of slots in slot arrays
tween the corrugated surface and the PEC plate on to avoid grating lobes. Four independent quasi-
top is d = 2 mm. TEM waves can be observed. It is worth recalling
that, unlike a parallel plate waveguide, where a
linear phase-front would inevitably lead to linear
amplitude patterns in transverse cuts, the single
hard wall waveguide amplitude pattern seen in
Fig. 7a remains uniform. The preservation of the
phase profile is also shown in Fig. 7b.

Figure 7. Magnetic field produced by four probes fed with uniform magni-
tude and a linear phase of 45º. Two transverse cuts, at 4λ and 8λ away from the
source, are shown.

It is worth noticing that the corrugated surface

should not be seen as a mere choke to the trans-
verse propagation. If that was the case, deep, Figure 8. Drawing of the hard surface consisting
air-filled grooves could have been used. The fact of narrow strips and via-holes connecting the strips
is that air-filled grooves do not work and the rea- to the ground plane.
son was stated above: the corrugated surface
must be simultaneously soft and hard. In other
words, the frequency band for which the corru- 2.2. Experimental demonstration
gated surface functions in hard mode coincides A practical realization has been manufactured
with the soft mode band of the surface, and that and measured to show the remarkable proper-
only happens when grooves are filled with a di- ties observed previously in simulations. As stat-
electric material. ed above, longitudinal grooves must be filled
with a dielectric material of permittivity , in
Additional understanding comes from the ob- order to work as a hard surface, and the groove
servation of the dispersion diagram obtained depth should be approximately at the working
for the hard direction of a single hard wall frequency. From a manufacturing point of view,
waveguide with metal sidewalls (Fig. 6). it is easier to construct the edges between cor-
rugations by perforating a copper-clad laminate
A set of quasi-TEM eigenmodes can be seen. At with a series of via-holes than machining a cor-
the intersection point, all the eigenmodes share rugated surface and filling the grooves up with
the same phase velocity, and the propagation the desired material (see Fig. 8). This is the proce-
constant is almost that one of light in vacuum. dure followed here to construct the hard surface
That point establishes the hard condition. As at the expense of experiencing some deviation
explained above, local waves are obtained at from the expected theoretical performance of a
this point as a linear combination of these ei- corrugated hard surface.
genmodes. Interestingly, when corrugations are
air-filled there is no intersection point. Therefore Fig. 9 shows the hard surface made out of metal-
eigenmodes have different phase velocities and lic strips etched on a copper-clad laminate, and
the sought local wave cannot be obtained. Si- via-holes connecting the strips to the ground
multaneously, a wide stopband exists in the soft plane. A detail of the hard surface is also shown

68 ISSN 1889-8297 / Waves · 2011 · year 3

 Gap waveguide
are
ad­vantageous
relative to
existing high-
frequency
transmission
lines, in terms
of losses, cost,
bandwidth,
packaging and
Figure 9. Hard surface made out of etched copper strips and plated via-holes on a PCB. Detail showing the integration.
coaxial probe.

in the picture where the coaxial probe used to cially antennas for millimeter- and submillimiter-
feed the waveguide can also be seen. The probe wave frequencies requiring control of the beam
is placed a quarter wavelength away from the steering, for instance for satellite mobile com-
short-circuit. munications.

A conducting surface is 2 mm above the hard A preliminary design of a 1-D slot array on a
surface. The waveguide is short-circuited at 5 SHWW for linear polarization is shown in Fig. 11.
mm from the receiving probe and terminated Eight transverse slots excited by a single probe
with an absorbing material. Both, left and right are considered. The waveguide is filled with a
sides are left open. The electric field pattern was dielectric material of permittivity 1.5 to reduce
measured in the waveguide section at a given grating lobes. Slots are equally spaced apart one
frequency. Cuts measured at several sections waveguide wavelength (λg). The radiation pat-
along the waveguide will show divergence of the tern pointing at broadside direction can be seen.
local wave. Fig. 10 shows the patterns obtained
for three cuts taken at 40, 100, and 160mm (2λ, A complete design of a circularly polarized 2-D
5λ, and 10λ) away from the source in longitudi- array was done. Given that the unique propa-
nal direction. Samples were taken at the points gation allowed inside a SHW-waveguide is of
shown with symbols. The field obtained from quasi-TEM type, circularly-polarized waves can
simulations on a SHWW whose hard surface use be generated by a radiating element constituted
continuous grooves is also shown for reference. by a pair of slots oriented at ±45º which are λg/4
As can be seen, the pattern is very well preserved apart. First, a row of slots, i.e. a 1-D array, was de-
with distance. The beam is certainly very narrow: signed. Later, the 2-D array was obtained just by
levels are more than 10 dB below the maximum replicating that row of slots. This was possible
5 mm (λ/4) away from it. All values are normal- due to the insignificant mutual coupling among
ized with respect to the first cut maximum. rows of slots on a SHWW, unlike an oversized rec-
tangular waveguide where the coupling among
rows of slots cannot be neglected.
0
40 mm
−5
100 mm
160 mm
For the 1-D array design, an optimization proc-
CST ess is always needed to fulfill antenna specifi-
−10
cations. Magnitude and phase of the array ele-
S21 (dB)

−15 ments must be tuned up by adjusting element

position and slot length. In our case, the design
−20
is broadside and for maximum efficiency. Be-
−25 sides, position of each slot of a pair must be ad-
justed to get circular polarization. The algorithm
−30
−20 −15 −10 −5 0 5 10 15 20
Width (mm)

Figure 10. S21 parameter measured for three

transverse cuts taken at 40, 100, and 160mm (2λ,
5λ, and 10λ) away from the source in longitudinal
direction. A CST simulation is shown for reference.

2.3. Application to slot array antennas

The capability of a SHWW to channel several
independent local quasi-TEM waves lend it a Figure 11. 8-element linear array of transverse slots on a SHWW: (a) sketch,
promising potential to feed slot arrays, espe- (b) simulated radiation pattern.

Waves · 2011 · year 3 / ISSN 1889-8297 69

 The previous idea was also inspiration for a
new metamaterial-based waveguide technol-
ogy especially appropriate for millimeter and
sub-millimeter waves. This new technology,
so-called ridge gap waveguide (RGW), was first
introduced in [9]-[11]. This waveguide is made
between two parallel metal plates. One of them
Figure 12. Sketch of a circularly polarized 2-D slot is made of a texture to create a high impedance
array of 4 × 10 elements. condition at the surface (ideally, a PMC bound-
ary), and therefore to impose a cut-off for the
parallel-plate modes. In between the textured
proceeds in an iterative fashion adjusting loca-
plate, there is a metal ridge to guide the wave
tion and size of the slots to accomplish desired
along a particular path. Thereby, the waveguide
magnitude and phase of the fields radiated by
is formed in the gap between the ridge and the
the array elements. The iterative procedure ends
metal plate on top, as illustrated in Fig. 15.
when a convergence criterion is reached. Once
the 1-D array has been obtained, it is replicated
Both waveguiding structures, the SHWW and
to give the 2-D array of Fig. 12. Rows are 0.75λ
the RGW, can guide local quasi-TEM waves.
equally spaced apart to avoid grating lobes. Two
However, the waves are forced to follow straight
cuts of the radiation pattern in longitudinal and
paths within the SHWW since they must go par-
transverse directions are shown in Fig. 13.
allel to the corrugations. Whereas, for the RGW
the waves have to follow metal ridges, which
can bend, split, make corners, etc. Therefore,
the RGW is a more versatile structure than the
SHWW to make high-frequency circuits.

The key factors of this new waveguide technol-

ogy can be enumerated as follows:

• low losses, since gap waveguides do not re-

quire dielectrics and they can be completely
realized with metal,

• low cost, as there is no need for any con-

Figure 13. Radiation pattern of the circularly polarized 2-D slot array: (a) longi-
ducting joint between the two plates, what
tudinal cut, (b) transverse cut.
makes the manufacturing easier,

• large bandwidth, indeed a bandwidth larger

3. Ridge gap waveguides than one octave appears to be feasible [16],
3.1. Structure and operation
• no packaging trouble, because gap
The single hard wall waveguide (SHWW) was
waveguides can be designed to be com-
originally conceived to feed a slot array. The
pletely enclosed, without the presence of
SHWW consists of an oversized rectangular
cavity modes, which also makes it an attrac-
metal waveguide or parallel-plate waveguide
tive packaging technology for microstrip cir-
with the bottom metal plate replaced by a hard
cuits [17],
surface. Another interpretation of using a hard
surface as the bottom surface comes from the
• easy integration of active components like
observation of how fields propagate within a
amplifiers and MMICs because shielding and
parallel-plate waveguide (PPW) (see Fig. 14). On
packaging are more or less automatically
the one hand, a vertical electric field propagates
provided by the gap waveguide itself.
freely between two parallel metal plates or PEC
plates regardless of the distance between them.
All these factors make the gap waveguide ad-
On the other, all modes are below cut-off be-
vantageous relative to existing high-frequency
tween a PEC plate and a PMC plate provided the
transmission lines, such as microstrip or copla-
distance between the plates is smaller than λ/4.
nar waveguides, and hollow metal waveguides.
Thus, by using a hard surface (ideally, a PEC/PMC
On one hand, microstrip or coplanar waveguides
strip grating) and a parallel PEC plate separated
need dielectrics and in consequence suffer from
by an air gap smaller than λ/4, it is possible to
losses with increasing frequency; they have lim-
combine both performances depending on the
ited power handling capability, and spurious res-
direction of propagation, so that this waveguide
onances when encapsulated. On the other, con-
works like a PEC-over-PEC PPW in the longitudi-
ventional hollow waveguides are realized in two
nal direction, favoring vertically polarized waves,
parts which have to be joined together, what in-
whereas it works like a PEC-over-PMC PPW (with
creases its manufacturing cost, especially at high
h<λ/4) in the direction transverse to the strips
frequencies when the sidewalls become thinner
suppressing any kind of propagation.
and thinner. Thus, this new technology seems

70 ISSN 1889-8297 / Waves · 2011 · year 3

 Figure 14. Field propagation in (a) PEC-over-PEC PPW, (b) PEC-over-PMC PPW.

to be an interesting transmission line for realiz- of the pins should be approximately λ/4 to trans-
ing components and circuits from a frequency form the ground plane (PEC plate) to a high im-
around 30 GHz, when the current technologies pedance surface (ideally, a PMC plate).
already exhibit some deficiencies, up to 100 GHz.
The basic geometry of the ridge gap waveguide The characteristics of the ridge gap waveguide
comprises two parallel conducting surfaces. One can be seen from its dispersion diagram. The
of the surfaces is provided with a texture that is diagram shows a set of rectangular waveguide
used to realize conducting ridges surrounded by type modes appearing below 10 GHz. They have
a high impedance surface. The new waveguide a lower cut-off similar to normal rectangular
is located in the gap between metal surfaces. waveguide modes, but go into a stopband at
The gap is usually filled with air, but it can also 10 GHz approximately, and appear again at the
be dielectric-filled, and its size should be smaller end of this stopband, around 20 GHz. Inside the
than λ/4. The textured surface stops waves in all whole stopband there is only one mode very
directions (provided the waveguide height is close to the light line. This is the desired local
low enough) in such a way that the waves have quasi-TEM mode following one ridge. Therefore,
to follow the metal ridges. it can be stated that the bandwidth of the ridge
gap waveguide, i.e. the frequency band where
the waveguide works as expected, is nearly the
unimodal band within which only one quasi-
TEM mode is present.

Figure 15. Cross section of a canonical ridge gap

waveguide, with a ridge surrounded by a high imped-
ance surface.

A metal parallel-plate waveguide with h<λ/2

can only guide vertically polarized TEM waves
in any direction without any low-frequency
cut-off. These TEM waves span throughout the Figure 16. Geometry of the ridge gap waveguide
whole waveguide, thus they are global parallel- in bed of nails.
plate modes. Whereas, the quasi-TEM waves of a
ridge gap waveguide are local waves following
one ridge. If the global parallel-plate modes are
present in this waveguide, they will destroy the 400
Propagation constant (m−1 )

local gap waveguide performance completely. e

350 lin
Therefore, it is crucial for the performance of ht
300 lig
the ridge gap waveguide that all kinds of global
250
parallel-plate modes are prohibited from propa-
200
gating. This is achieved by texturing the metal
150
surface on both sides of the ridge in such a way
that it gets high surface impedance and pro- 100

vides cut-off for the global parallel-plate modes 50

within a certain frequency band. A bed of nails 0
0 5 10 15 20 25
or a Fakir's bed has been chosen to realize the Frequency (GHz)
high impedance surface because it can mimic an
ideal impedance boundary [8]. This bed of nails Figure 17. Dispersion diagram in the ridge direc-
is formed by metal pins standing vertically over tion of a RGW in bed of nails.
a ground plane (see Fig. 16). The depth or height

Waves · 2011 · year 3 / ISSN 1889-8297 71

 3.2. Application to microwave circuits Once a RGW in bed of nails has been character-
To be able to design circuits in a new waveguide ized, we can design circuits in RGW in the same
technology is essential to characterize it as a way as we do it in microstrip technology, for
transmission line and therefore to know its prop- instance. Some microwave circuits in ridge gap
agation constant and characteristic impedance. waveguide technology, in particular a power
The similarity between the ideal model of a RGW splitter and a coupler have been designed,
and the stripline, as it is shown in Fig. 17, allow manufactured and measured. Numerical results
using the formulas for the characteristic imped- as well as measurements demonstrate good
ance of the latest as a first approximation to the performance within a wide frequency band.
characteristic impedance of a RGW. It is easy to Even though this technology is particularly in-
see that the gap waveguide is a sort of half strip- teresting for millimeter and submillimeter-wave
line. Nevertheless, this approximation is not ac- bands, initial developments have been done in
curate enough for real cases, and we have to re- the microwave band for the sake of simplicity
sort to numerical approaches as referred in [18]. during the manufacturing stage.

• A power divider at 15 GHz was designed

and manufactured. Fig. 18a shows the man-
ufactured prototype. Measurements were
made with a Network Analyzer using an APC-
3.5mm calibration kit. Good agreement be-
tween HFSS simulations and measurements
can be observed at 15 GHz in Fig. 18b. Return
loss are lower than -15 dB while insertion loss
is around 4 dB.
Figure 18. (a) Ideal gap waveguide and (b) stripline models.
• A typical branch line coupler was designed
at 15 GHz. Fig. 19a shows a photo of the con-
structed prototype. Fig. 19b shows simulated
total electric field at the top of the ridge. The
directional aptitude can be noted.

The magnitude of S parameters are shown in

Fig. 21.a where a good agreement between
simulations and measurements can be observed
around 15 GHz. Return loss are lower than -24
dB at 15 GHz while insertion loss in the coupled
branches are 2.2 dB and 4.2 dB. A bandwidth of
Figure 19. (a) Prototype of the power divider in RGW; (b) simulated (dashed 3% has been measured with a threshold of -10
lines) and measured (continuous lines) S parameters dB in S11 parameter. In this bandwidth a phase
difference between 81º and 89º has been meas-
ured, as it is shown in Fig. 21.b.

Some microwave circuits have been designed in

ridge gap waveguide technology showing good
performance not only in simulations but also in
measurements. These circuits can be useful to
design more complex circuits, like microwave
filters and corporate feeding networks for array
antennas, in future.

Figure 20. (a) Prototype of the branch line coupler in RGW; (b) electric field at
15 GHz. 4. Conclusions
Gap waveguide technology is a promising pro-
posal for easy integration of active and passive
components and may eventually replace con-
ventional packaging techniques. This technol-
ogy is still in its infancy and has to go a long way
to demonstrate its feasibility. Therefore, more
work needs to be done regarding actual losses,
transitions to other technologies and simpli-
fied models to ease circuit design. In this review
paper, the physical principles and some experi-
mental demonstrations have been provided.
Figure 21. (a) Simulated (dashed lines) and measured (continuous lines) S pa-
rameters; (b) coupled branch phase difference.

72 ISSN 1889-8297 / Waves · 2011 · year 3

 Acknowledgements waveguide with a hard surface on one broad
wall for planar slot array antenna applica-
This work has been supported by the Spanish tions,” in Antennas and Propagation, 2009.
Ministry of Science and Technology (Ministe- EuCAP 2009. 3rd European Conference on,
rio de Ciencia y Tecnología) under the CICYT Mar. 2009, pp. 41-44.
projects TEC2010-20841-C04-01 and CSD2008- [13] Z. Sipus, M. Bosiljevac, and P.-S. Kildal, “Lo-
00068. cal wave Green's functions of parallel plate
metamaterial-based gap waveguides with
one hard wall,” in Antennas and Propagation,
References 2009. EuCAP 2009. 3rd European Conference
on, Mar. 2009, pp. 37-40.
[1] M. Samardzija, J. Hirokawa, and M. Ando, [14] P.-S. Kildal, “Artificially soft and hard surfaces
“Dominant quasi-TEM mode generator for in electromagnetics,” Antennas and Propa-
thin oversized dielectric coated rectangular gation, IEEE Transactions on, vol. 38, no. 10,
waveguide,” in Antennas and Propagation pp. 1537-1544, Oct. 1990.
Society International Symposium 2007, IEEE, [15] P. Kildal and A. Kishk, “EM Modeling of sur-
June 2007, pp. 4272-4275. faces with STOP or GO characteristics - artifi-
[2] J. Fernández and M. Sierra-Castañer, “Elec- cial magnetic conductors and soft and hard
tromagnetic bandgap structures as artificial surfaces,” ACES Journal, vol. 18, no. 1, pp. 32-
magnetic conductor surfaces sidewalls in 40, Mar. 2003.
parallel plate slot antennas,” Microwave and [16] E. Alfonso, P.-S. Kildal, A. Valero-Nogueira,
optical technology letters, vol. 48, no. 7, pp. and J. Herranz, “Numerical analysis of a met-
1441-1446, 2006. amaterial-based ridge gap waveguide with
[3] C. Balanis, Advanced engineering electro- a bed of nails as parallel-plate mode killer,”
magnetics. Wiley New York, 1989. in Antennas and Propagation, 2009. EuCAP
[4] E. Alfonso, A. Valero-Nogueira, J. Herranz, 2009. 3rd European Conference on, Mar.
and D. Sanchez, “Oversized waveguides for 2009, pp. 23-27.
TEM propagation using hard surfaces,” in [17] E. Rajo-Iglesias, A. Zaman, and P.-S. Kildal,
Antennas and Propagation Society Interna- “Parallel plate cavity mode suppression in
tional Symposium 2006, IEEE, July 2006, pp. microstrip circuit packages using a lid of
1193-1196. nails,” Microwave and Wireless Components
[5] A. Valero-Nogueira, E. Alfonso, J. Herranz, Letters, IEEE, vol. 20, no. 1, pp. 31-33, Jan.
and M. Baquero, “Planar slot array antenna 2010.
fed by an oversized quasi-TEM waveguide,” [18] E. Alfonso, P.-S. Kildal, A. Valero-Nogueira,
Microwave and Optical Technology Letters, and M. Baquero, “Study of the characteris-
vol. 49, no. 8, pp. 1875-1877, 2007. tic impedance of a ridge gap waveguide,” in
[6] A. Valero-Nogueira, E. Alfonso, and J. Her- Antennas and Propagation Society Interna-
ranz, “Multi-beam waveguide for quasi-TEM tional Symposium, 2009. APSURSI '09. IEEE,
modes,” Spanish Patent No. P200803558, De- June 2009, pp. 1-4.
cember 2008.
[7] P.-S. Kildal, “Definition of artificially soft and
hard surfaces for electromagnetic waves,” Biographies
Electronics Letters, vol. 24, no. 3, pp. 168-
170, Apr. 1988.
[8] M. Silveirinha, C. Fernandes, and J. Costa, Esperanza Alfonso Alós
“Electromagnetic characterization of tex- received the degree in
tured surfaces formed by metallic pins,” An- electrical engineering
tennas and Propagation, IEEE Transactions from the Universidad
on, vol. 56, no. 2, pp. 405-415, Feb. 2008. Politécnica de Valencia
[9] P.-S. Kildal, E. Alfonso, A. Valero-Nogueira, (UPV), Valencia, Spain, in
and E. Rajo-Iglesias, “Local metamaterial- 2004, and the Ph.D. in
based waveguides in gaps between parallel electrical engineering
metal plates,” Antennas and Wireless Propa- also at UPV in 2011. She
gation Letters, IEEE, vol. 8, pp. 84-87, 2009. has been with the Institute of Telecommunica-
[10] P.-S. Kildal, “Waveguides and transmission tions and Multimedia Applications of the UPV
lines in gaps between parallelconduct- since 2004 to 2011. She is currently working as a
ing surfaces,” European patent application Pos-Doc for Chalmers University of Technology,
EP08159791.6, July 2008. Sweden. Her main research interests include
[11] P.-S. Kildal, “Three metamaterial-based gap analysis and design of slot array antennas, nu-
waveguides between parallel metal plates merical methods, millimiter and submillimiter
for mm/submm waves,” in Antennas and waveguide technology and metamaterials.
Propagation, 2009. EuCAP 2009. 3rd Euro-
pean Conference on, Mar. 2009, pp. 28-32.
[12] S. Skobelev and P.-S. Kildal, “Analysis of glo-
bal eigenmodes in an oversized rectangular

Waves · 2011 · year 3 / ISSN 1889-8297 73

 Alejandro Valero-Nogue- with the Communications Department and the
ira was born in Madrid, Institute of Telecommunications and Multimedia
Spain on July 19, 1965. He Applications, UPV. His main research interests in-
received the M.S. degree clude microwave circuit and antenna analysis,
in electrical engineering design and measurement.
from Universidad Politéc-
nica de Madrid, Madrid,
Spain in 1991 and the Miguel Ferrando-Bataller
Ph.D. degree in electrical was born in Alcoy, Spain,
engineering from Universidad Politécnica de Va- in 1954. He received the
lencia, Valencia, Spain in 1997. In 1992 he joined M.S. and Ph.D. degrees
the Departamento de Comunicaciones, Univer- in electrical engineering
sidad Politécnica de Valencia, where he is cur- from the Universidad Po-
rently an Associate Professor. During 1999 he litecnica de Cataluña, Bar-
was on leave at the ElectroScience Laboratory, celona, Spain, in 1977 and
The Ohio State University, Columbus, where he 1982, respectively. From
was involved in fast solution methods in electro- 1977 to 1982, he was a Teaching Assistant with
magnetics and conformal antenna arrays. His the Antennas, Microwave, and Radar Group, Uni-
current research interests include computation- versidad Politécnica de Cataluña, and in 1982 he
al electromagnetics, Green’s functions, became an Associate Professor.
waveguide slot arrays, millimetre-wave technol-
ogy and automated antenna design procedures. In 1990, he joined Universidad Politécnica de Va-
lencia, Valencia, Spain, where he is a Professor. His
current research activities include numerical meth-
Jose I. Herranz-Herruzo ods, antenna design, and e-learning activities.
was born in Valencia in
1978. He received the M.S.
degree in electrical engi- Vicent Miquel Rodrigo
neering from Universidad Peñarrocha was born in
Politécnica de Valencia, Valencia, Spain, on Sep-
Spain in 2002. He is cur- tember 29, 1966. He re-
rently working toward ceived the Ingeniero de
the Ph.D. degree. Since Telecomunicación degree
2002 he is with the Departamento de Comuni- from Universidad Politéc-
caciones at Universidad Politécnica de Valencia, nica de Madrid (UPM),
where he is currently an Assistant Professor. His Madrid, Spain, in 1990,
main research interests include slot array anten- and the Ph.D. degree in electrical engineering
na design and optimization and efficient com- from the Universidad Politecnica de Valencia in
putational methods for printed structures. 2003. He joined the Departamento de Comuni-
caciones at Universidad Politécnica de Valencia
in 1991 as a Lecturer. His current interests in-
Mariano Baquero clude radio-wave propagation over the sea, in-
was born in Murcia, Spain, strumentation, and any educational activity.
on January 11, 1962. He
received the degree in tel-
ecommunications engi- Felipe Vico
neering from Universidad was born in Valencia,
Politécnica de Cataluña Spain, in 1981. He re-
(UPC), Barcelona, Spain, ceived the M.S degree in
in 1986 and the Ph.D. 2004 in Telecommunica-
degree from Universidad Politécnica Valencia tion engineering from
(UPV), Valencia, Spain, in 1994. He became a the Polytechnic Univer-
Member (M) of IEEE in 1987. He was with the An- sity of Valencia and the
tennas, Microwave and Radar Group, UPC, from Ph.D. degree in 2009 in
1986 to1988, where he worked on the develop- Telecommunications engineering also from UPV.
ment of a cylindrical near-field facility to meas- He received the M.S in Mathematics in 2009 from
ure a 3-D radar antenna in CESELSA. Since 1989, Universidad Nacional de Educación a Distancia.
he has been with the UPV where he became a He received the best paper award in EUCAP
Full Professor in 2003. During 1995, he held a 2006. He did a research stay in 2010 in New York
postdoctoral grant at the Joint Research Cen- University with professor Leslie Greengard in
tre, European Commission, Ispra, Italy, where he integral equation methods. He is now assistant
developed high-resolution algorithms for radar Professor in Universidad Politécnica de Valencia.
applications. From April 1996 to February 1998, His research interests are Numerical methods
he was a Vice-Dean of the Telecommunications for CEM, high frequency methods and integral
Engineering School of Valencia. He is currently equation methods.

74 ISSN 1889-8297 / Waves · 2011 · year 3

 Eva Antonino-Daviu Daniel Sánchez-Escuderos
was born in Valencia, was born in Vila-real,
Spain, on July 10, 1978. Spain on October 20,
She received the M.S. and 1980. He received the M.S.
Ph.D. degrees in electrical and PhD. degree in elec­
engineering from Uni- trical engineering from
versidad Politécnica de Universidad Politécnica
Valencia, Valencia, Spain, de Valencia, in Valencia,
in 2002 and 2008, respec- Spain, in 2007 and 2009,
tively. In 2002 she joined the Electromagnetic respectively. He has been with the Institute of
Radiation Group, Universidad Politécnica de Telecommunications and Multimedia Applica-
Valencia, and in 2005 she became a Lecturer at tion of Universidad Politécnica de Valencia since
the Escuela Politécnica Superior de Gandia, Gan- 2005. His main re­search interest includes near
dia, Spain. During 2005 she stayed for several to far field transfor­mation, antenna diagnostics
months as a guest researcher at the Department and, THz technology and applications.
of Antennas & EM Modelling of IMST GmbH, in
Kamp-Lintfort, Germany. Her current research
interests include wideband and multi-band Bernardo Bernardo Cle-
planar antenna design and optimization and mente was born in Valen-
computational methods for printed structures. cia, Spain, on May 8, 1972.
Dr. Antonino-Daviu was awarded the “Premio He received the degree in
Extraordinario de Tesis Doctoral” from the Uni- electrical engineering
versidad Politécnica de Valencia in 2008. from Universidad Politéc-
nica de Valencia (UPV),
Valencia, Spain, in 2003,
Marta Cabedo-Fabrés and He is currently work-
was born in Valencia, ing toward the Ph.D. degree at Universidad
Spain, on June 8, 1976. Politécnica de Valencia. He has been with the In-
She received the M.S. and stitute of Telecommunications and Multimedia
Ph.D. degrees in electrical Application of Universidad Politécnica de Valen-
engineering from Uni- cia since 2005. His main research interests in-
versidad Politécnica de clude near field antenna measurement.
Valencia, Spain, in 2001
and 2007, respectively. In
2001, she joined the Electromagnetic Radiation Antonio Vila Jiménez
Group at Universidad Politécnica de Valencia was born in Valencia,
(UPV), as a Research Assistant. In 2004, she be- Spain on June 18, 1981.
came a Lecturer in the Communications Depart- He received the degree
ment, Universidad Politécnica de Valencia. Her in Telecommunications
current scientific interests include numerical Engineering, specialising
methods for solving electromagnetic problems, in Telecommunication
and design and optimization techniques for Systems from Universidad
wideband antennas for mobile terminals, and Politécnica de Valencia,
MIMO antennas for Personal Area Networks. Valencia, Spain, in 2007. He has been with the
Institute of Telecommunications and Multime-
dia Application of Universidad Politécnica de Va-
lencia since 2007. His main research interest in-
cludes field antenna measurement and antenna
fabrication.

Waves · 2011 · year 3 / ISSN 1889-8297 75