A Novel Differential Microstrip Patch Antenna and

Array at 79 GHz
#Z. Tong1, A. Stelzer1,2, C. Wagner2, R. Feger2, E. Kolmhofer3
1
Institute for Communications & Information Engineering, Johannes Kepler University
Altenberger Str. 69, A-4040 Linz, Austria, {z.tong, c.wagner, r.feger, a.stelzer}@icie.jku.at
2
Christian Doppler Laboratory for Integrated Radar Sensors, Johannes Kepler University
Altenberger Str. 69, A-4040 Linz Austria
3
DICE GmbH & Co KG
Freistaedter Strasse 400, 4040 Linz, Austria, [email protected]

Abstract
A novel differential microstrip patch antenna (DMPA) is designed for autonomous cruise control
radar systems at 79 GHz. Distinct from conventional single-ended patch antennas, the DMPA uses a pair
of coupled lines as feeding line. It eliminates the need of a balun in the RF frontend and supports the
realization of more compact radar frontend. First, a single DMPA is designed. The impedance bandwidth
reaches 4.7 GHz and the gain at 79 GHz is 6.2 dBi. Simulations show that the length of the patch
determines the resonance frequency and the gap between the feed points affects the patch impedance.
Furthermore, a four-element series-fed array is presented with 4.6 GHz bandwidth and 12.8 dBi gain.
Both the single DMPA and the four-element DMPA array were fabricated and measured. Measurement
results show good agreement with simulation results.

1. Introduction
Differential circuits are used in silicon MMICs [1]. Unfortunately most of the conventional
antennas are of single-ended type. Therefore at least a balun is needed between the differential circuits and
the single-ended antenna/array. Research on differential antennas has become more and more attractive
recently [2, 3]. These differential antennas use either proximity coupled structures or probe feed structures
and frequency ranges are below 10 GHz. In this paper, a novel differential microstrip patch
antenna/array (DMPA) for autonomous cruise control (ACC) applications at 79 GHz is presented. For this
frequency range, to the authors’ knowledge, it is the first time that a fully differential microstrip
antenna/array is presented. The novel DMPA eliminates not only the loss dure to a balun, it also inherits
significant advantages from the microstrip line structures, such as low profile, compact size, low cost,
simple fabrication, etc.

Fig. 1. Structure of a single DMPA and the E-field. Fig. 2. A four-element series-fed DMPA array.
The structure of a single DMPA and the E-field are shown in Fig. 1. It consists of the radiation
part-antenna patch and the feeding part-coupled lines. Through the feeding line, the differential signals are
injected into the non-radiant edge of the antenna patch (left side of Fig. 1, the positive port is denoted by
P+, the negative port by P-). The electric field of the patch is shown on the right side of Fig. 1. The Z
direction is perpendicular to the antenna plane. The length of the patch determines the resonance
frequency of the patch, the distance between the feeding points (centers of the coupled lines) determines
the impedance of the patch. A more detailed discussion on this topic is given in Section 2.
At first, a single DMPA is presented. Later on, a four-element series-fed DMPA array was also
designed (see Fig. 2). Four antenna patches are connected by the coupled lines. A matching network at the
first patch was used for the impedance match of the antenna array. Both antenna and array were fabricated
on Taconic TLE-95 and measured.

2. Design Procedure
In this section the design rules for a single DMPA and a four-element series-fed DMPA array at
79 GHz are discussed. The design procedure of the single DMPA includes two steps. In the first step, the
dimension of the patch can be calculated according to [4]. The width of a single patch is given by
c 2
W= (1)
2 fr εr +1

with c as the free-space velocity of light, εr ( = 3) as the relative permittivity, and fr as the resonance
frequency. The length L of a single patch is calculated to
1
L= λr − 2Δl (2)
2
⎛W ⎞
(ε eff + 0.3)⎜ + 0.264 ⎟
Δl ⎝ h ⎠ (3)
= 0.412
h ⎛W ⎞
(ε eff − 0.258)⎜ + 0.8 ⎟
⎝h ⎠

with εeff as the effective permittivity, λr as the relative wavelength. They can be calculated according to [4].
In (3), h ( = 0.127 mm) is the height of the substrate. The patch parameters are calculated as following:
L1 = 0.98 mm and W1 = 1.34 mm. The loci of the patch impedance in the Smith chart is approximately a
circle with constant conductance (see Fig. 3), but the center is shifted to the upper part. The patch
impedance is determined by the feeding positions. With increasing the spacing (sn) between the feeding
points, the patch impedance at 79 GHz increases rapidly (see Fig. 3). If the feeding points are located too
close to one another (sn < 0.1 λ0), the patch resonance will disappear. Therefore, the second step is to
optimize the patch impedance. With a 3D EM simulator, a single DMPA was design for 100 Ohm
differential impedance. Simulation results show an impedance bandwidth of 3.1 GHz and 6.1 dBi antenna
gain at a frequency of 79 GHz.

Fig. 3. Patch impedance with different feeding points distances. Fig. 4. Matching network function.
 Starting from the dimension of a single DMPA, it is possible to extend to a four-element series-
fed array (see Fig. 2). The antenna array is composed of four antenna patches, which are connected by a
pair of coupled lines in between. The design of an antenna array is more complicated than of a single
patch antenna. Besides the impedance match, the radiation pattern is another important factor to be
considered. The distance between the antenna patches (center to center) is a critical parameter influencing
the radiation pattern. It equals the sum of the width of the patch (W) and the gap between the antenna
patches (d0). In order to keep each antenna patch radiating in phase at the center frequency, a distance of
3.48 mm (0.92 λ0) is used.
Compared to the single DMPA, the series-fed DMPA array introduces more inductance in the
antenna impedance. A matching network is designed between antenna patches and feed line. The
matching network consists of an inset on the first patch and quarter wavelength coupled lines. The inset
works as a section of transmission line, it converts the antenna impedance from inductive part in Smith
Chart (curve 1 in Fig. 4) to the high resistance part of real axis (curve 2). Then the quarter wavelength
transforms the antenna impedance to 100 Ohm differential impedance.
Based on simulation, a four-element series-fed antenna array is realized with following
dimensions: d0 = 2.08 mm, L2 = 1.1 mm, L3 = 0.96 mm and W2 = W3 = 1.4 mm. The total length of the
array is 11.84 mm (without the matching network). The impedance bandwidth, derived from simulation
results, is 4.1 GHz and the antenna gain is 12.0 dBi, respectively.

3. Measurement
Both, the single patch antenna as well as the array, were fabricated and measured. For the S-
parameter measurement, it is difficult to get one-port differential S-parameters directly. Therefore, the
antenna/array is measured as a single-ended two-port device at first. Then, the differential S-parameters
are calculated by using the following matrix conversion [5]:
1 ⎡1 − 1⎤
M= ⎢ ⎥
(4)
2 ⎣1 1⎦

S dm = M⋅ S ⋅M (5)

with S as the measured single-ended two-port S-parameters and Sdm as the calculated one-port differential
S-parameters.

(a) (b)
Fig. 5. Measured and simulated return loss of a single DMPA (a) and of a four-element DMPA array (b).
Figure 5 shows the measurements in comparison with the simulation results of the return loss of
the two antennas. The measured bandwidth of a single DMPA is 4.7 GHz, that of a four-element series-
fed DMPA array is 4.6 GHz respectively. The bandwidths of the measurement results are wider than the
simulation results. This is possible because of the losses in the substrates.
During the far field measurements, a waveguide to differential microstrip line transition was used
(see Fig. 6). The waveguide signal is injected from the back of the transition and converted to a
differential signal through the transition. A short end cap with an open channel is placed on the top of the
transition. The differential antenna/array can be connected to the output of the transition directly. The
measurement setup is shown in Fig. 7.

Fig. 6. Photograph of the transition and Fig. 7. Photograph of the far field measurement
four-element series-fed DMPA array. setup.
The normalized E-plane and H-plane, co-polarization and cross-polarization radiation patterns of
both antennas are plotted in Fig. 8 and Fig. 9. The simulation results of the cross-polarization in H-plane
are neglected here because of its quite low level.
There were additional noises observed within the far field measurements. Therefore average
values (20 samples per degree) were used for plotting the results. Because of the height of the transition
cap, the last several degrees of the radiation patterns in the H-plane (85~90 degree) are disturbed.
Within the far field measurements, some ripples were observed in the radiation pattern, stronger in
the E-plane than in the H-plane, stronger for the single patch antenna than the antenna array. Radiation
from the surface wave at the edges of PCBs could be the most likely explanation. Absorbing material was
used on the edges of PCBs. This reduced some surface wave effects, but also distorted radiation patterns at
the angle ranges ±80~90 degree.
The gain of the antenna/array at 79 GHz is also measured. The calibrated gain for the single
DMPA is 6.2 dBi and 12.8 dBi for the four-element array, respectively.

(a) (b)
Fig. 8. Normalized radiation pattern of a single DMPA for E-plane (a) and H-plane (b).
 (a) (b)
Fig. 9. Normalized radiation pattern of a four-element DMPA array of E-plane (a) and H-plane (b).

4. Conclusions

In this paper, a novel DMPA and a four-element series-fed DMPA array are presented. These
DMPAs may be integrated with differential circuits easily. They can support more compact RF frontend
size and reduce the loss of the balun. The general design rules of the single patch antenna and the antenna
array are illustrated also.
Both the single patch antenna as well as the antenna array were fabricated and measured. The
measured results agree well with simulated results.
Within far field measurements, radiation of surface waves is observed. Removing the influence
from the surface waves radiation is one of the goals for the further design.

Acknowledgments

The authors would like to thank Prof. Wolfgang Menzel from Ulm University for supporting far
field measurements facilities. The authors would also appreciate Mr. Ralf Rudersdorfer, Mr. Bernhard
Mayrhofer and Mr. Johann Katzenmayer from Johannes Kepler University Linz for their assistance in the
manufacturing process.

References
[1] C. Wagner, H. Forstner, A. Stelzer, and H. Jaeger, “A 79-GHz single-chip radar transceiver with
switchable TX and LO feedthrough in a silicon-germanium technology,” in IEEE Bipolar/BiCMOS
Circuits and Technology Meeting Monterey, CA, accepted for publication,2008.
[2] T. Brauner, R. Vogt, and W. Baechtold, “A differential active patch antenna element for array
applications,” IEEE Microwave and Wireless Components Letters, vol. 13, No 4, pp. 161-163, 2003.
[3] Y. P. Zhang, J. J. Wang, “Theory and analysis of differentially-driven microstrip antennas,” IEEE
Trans. On Antennas and Propagation, vol. 54, No. 4, 2006.
[4] Constantine A. Balanis, Antenna Theory: Analysis and Design, 2nd edition, John Wiley & Sons. Inc.,
USA, pp.727-730, 1997.
[5] David E. Bockelman, William R. Eisenstadt, “Combined differential and common-mode scattering
parameters: theory and simulation,” IEEE Tran. On Microwave Theory and Techniques, vol. 43, No. 7,
pp.1530-1539, 1995.