Conclusions: Chapter Eleven
Conclusions: Chapter Eleven
Conclusions: Chapter Eleven
CHAPTER ELEVEN
Conclusions
PETER S. HALL
School of Electronic and Electrical Engineering
The University of Birmingham
Edgbaston, Birmingham, UK
K. C. GUPTA
Department of Electrical and Computer Engineering
University of Colorado
Boulder, CO
11.1 INTRODUCTION
Chapter 1 set the scene for the following chapters by giving an introduction to the
topic of integrated circuitantenna modules, with a glossary of module types and an
overview of the design needs. It therefore seems appropriate to conclude the book
with a nal chapter than draws the many threads together by attempting to describe
the current capabilities and what now needs to be done. There are obvious dangers in
attempting such a description, since what we suggest is in the future may in fact be
happening now and our statements of limitations of methods may be contested by
their advocates. However, we feel that the benets of this chapter will outweigh the
problems and we ask the reader to take it in the spirit that it was intended.
410
11.2 OVERVIEW OF ANALYTICAL METHODS 411
11.2.1 Introduction
The development of analytical tools for integrated circuitantennas modules essen-
tially concerns the integration of tools for circuits and for antennas. Such tools have
been available for some time, but new developments are continulally being seen.
Some convergence of tools for antennas and passive circuits has been evident for
some time. For example, the method of moments is now regularly used for analysis
of both printed circuit antennas and circuits. The key issue is therefore the inclusion
of active devices into large passive structures.
There are many parallels that can be drawn concerning modeling of active devices
and passive structures:
Numerical techniques of the type described in this book and others [1] can be
applied to both. In active devices charge transport mechanisms must be
appropriately modeled to extract device performance in terms of geometry,
412 CONCLUSIONS
semiconductor properties, bias, and signal level. Similarly, with passive circuits
and antennas the aim is to nd the performance for a given geometry and
material properties.
Equivalent circuit representations can be made of both the active and passive
components. Such representations can be derived from measurements or from
the numerical simulations noted above.
simplied equivalent circuit models as in Chapter 7, although for small arrays the
nite-difference time-domain method has been shown to be useful (Chapter 6). The
unit-cell approach, dened in Chapter 8, uses accurate element modeling combined
with a Floquet mode method to represent an innite array. In that chapter, the
induced EMF method and method of moments is used. In principle, any method can
be used; the use of nite-difference time-domain combined with phase walls for a
passive array has recently been reported [2].
Element modeling can be performed with numerous methods and these are
compared in more detail in the next section. As noted above, these are all attempts to
include equivalent circuit device models into conventional methods for analyzing
passive electromagnetic structures.
In most integrated circuitantenna modules the device feature size is several
orders of magnitude smaller than the passive sections. Modeling can be done in a
number of ways:
1. Equivalent Circuit Method. Primarily developed for circuit design, the method
has been extended to antennas, where it is appropriate for geometries that
support either well-dened single-mode resonances, such as a thin microstrip
patch or thin slot, or for traveling wave antennas. It will not, in principle,
analyze arbitrary shaped antennas or circuit components. It is easy to
incorporate active devices where these are represented by equivalent circuits.
414 CONCLUSIONS
Good Ease of
Flexibility for Low Incorporating
Arbitrary High Com- Good Active
Method Chapter Structures Speed plexity Accuracy Device
ppp ppp p ppp
Equivalent circuit 3
p pp pp pp pp
Multiport network 4
p p ppp p
Integral equation 5
pp p ppp p
Finite-difference 6
time-domain
ppp ppp p
Finite element
pp p ppp p
Transmission line
matrix
p pp ppp
poor, moderate, good, very good.
It is fast and easy to implement but, in general, has good accuracy only for
well-characterized systems. Nonlinear performance can be obtained provided
the device model supports it.
2. Multiport network method (MNM). MNM is a network-based method that
allows analysis of a wider range of antenna shapes. More computation
(compared to the equivalent circuit method) is required with consequent
speed reduction. It will give better accuracy in radiation pattern prediction
than the equivalent circuit model.
3. Integral Equation Methods. These are computationally intensive methods with
relatively slow speed that will handle arbitrary shaped printed antennas and
dielectric layers. Finite ground planes and slabs can be dealt with only by
special additions. Accuracy can be excellent, but the addition of an active
device using equivalent circuit models is less straightforward, as care is needed
to derive appropriate voltages and currents at device terminals.
4. Finite-Difference Time-Domain Methods. These are in general simpler to
formulate than integral equation methods but again are computationally
intensive. Finite substrate and metallization can easily be included, as can
arbitrary shaped antennas and circuit components. Operation in the time
domain allows nonlinear effects such as intermodulation and harmonics to be
easily obtained. Care is needed with inclusion of active device equivalent
circuits.
5. Finite-Element Methods. Little work has been done on nite-element methods
for integrated circuitantenna modules. The nite-element method in the
frequency domain has recently been extended to include lumped elements
[7]. Operation in the time domain gives some advantages and Guillouard et al.
[8] describe the combination of the nite-element time-domain with the
SPICE method. The authors are aware of no comparative studies between
11.3 THE FUTURE 415
this method and other methods, although as shown in Table 11.2, it is expected
to have broad characteristics similar to the nite-difference time-domain
method.
6. Transmission Line Matrix (TLM) Method. The TLM method has widely been
used for a variety of electromagnetic problems, including structures with
lumped elements [9] and nonlinear elements [1013]. In principle, it can used
to model arbitrary shaped nite structures, but it is relatively complex to
implement and computationally intensive and hence slow. It can be accurate
and, because it is based in the time domain, can given the module's nonlinear
performance. Although eld components are calculated from the matrix
voltage and currents, care is still required in attaching the active device model.
with an awareness of the need that designers have for a variety of design tools,
ranging from fast, approximate methods to those that are highly accurate and
computationally intensive.
The current division between active device design and passive circuitantenna
design has been noted throughout this concluding chapter. The integration of
electromagnetic solvers with physical modeling allows some optimization of the
complete module, including active device processing parameters. This degree of
design integration could be very important for future MMIC circuitantennas
modules. However, bearing in mind the uctuations in device performance for a
given set of processing parameters, it is not clear at this stage just how signicant
this development might be.
Many of the applications suggested for integrated circuitantenna modules are in
the low cost, mass market personal communications and sensor areas. In some of
these applications, the effect of the local environment on the module may be
signicant. It may well be that the system will be set up so that it will not be
vulnerable to such perturbations. Nevertheless, there may well be a future demand
for analytical models that allow the effects of the local environment to be included.
Environmental effects on passive antennas, such as the interaction of the head and
hand on the operation of cellular telephone handsets, is now being widely
investigated with, for example, the nite-difference time-domain methods. With
the increasing importance of ``wearable'' communications devices such studies will
continue. If integrated circuitantenna modules are to have an impact on this market
sector, then analyses must be capable of including such effects.
This book has described analytical techniques for the design of integrated circuit
antenna modules. Such techniques have developed in the last decade and are now
of great signicance for the future progress of the technology. There are challenges
yet to be met and the area still remains a fertile one for analysts and engineers
alike.
REFERENCES
1. Itoh, T. (ed), Numerical Techniques for Microwave and Millimetre Wave Passive Struc-
tures, Wiley, New York, 1989.
2. Alexanian, A., Kolias, N. J., Compton, R. C., and York, R. A., ``Three dimensional FD-TD
analysis of quasi-optical arrays using Floquet boundary conditions and Berenger's PML,''
IEEE Microwave Guided Wave Lett., vol. 6, no. 3, Mar. 1996, pp. 138140.
3. Jacaboni, C., and Reggiani, L., ``The Monte Carlo method for the solution of charge
transport in semiconductors with applications to covalent materials,'' Rev. Mod. Phys., vol.
55, no. 3, 1983, p. 646.
4. Fischetti, M. V., and Laux, S. E., ``Monte Carlo simulation of submicron Si MOSFETs,'' in
Simulation of Semiconductor Devices and Processes, vol. 3, G. Baccarani and M. Rudan
(eds.), Pineridge Press, Swansea, UK, 1988, p. 349.
5. Hansch, W., ``The drift diffusion equation and its applications to MOSFET modelling,'' in
Computational Microelectronics, Springer-Verlag, New York, 1991.
REFERENCES 417
6. Ladbrooke, P. H., MMIC design GAAs FETs and HEMTs, Artech House, Boston, 1989.
7. Guillouard, K., Wong, F.-F., and Fouad Hanna, V., ``New global nite element analysis of
microwave circuits including lumped elements,'' IEEE MTT-S Symp. Dig., 1996, pp. 355
358.
8. Guillouard, K., Wong, M.-F., Fouad Hanna, V., and Citerne, J., ``New global time domain
electromagnetic simulator of microwave circuits including lumped elements based on
nite element method,'' IEEE MTT-S Symp. Dig., 1997, pp. 12391242.
9. Russer, P., Isele, B., Sohby, M., and Hosny, E. A., ``A general interface between YLM
models and lumped element circuit methods,'' IEEE MTT-S Symp. Dig., 1994, pp. 891
894.
10. Voelker, R. H., and Lomax, R. J., ``A nite difference transmission line matrix method
incorporating a non-linear device model,'' IEEE Trans. Microwave Theory Tech., vol. 38,
no. 3, 1990, pp. 302312.
11. Russer, P., So, P. P. M., and Hoefer, W. J. R., ``Modelling of non-linear active regions in
TLM method,'' IEEE Microwave Guided Wave Lett., vol. 1, no. 1, Jan. 1991, pp. 1013.
12. Hoefer, W. J. R., Isele, B., and Russer, P., ``Modelling of non-linear active devices in
TLM,'' IEE International Conference on Computation in Electromagnetics, London, Nov.
1991, pp. 2527.
13. Cascio, L., Tardick, G., and Hoefer, W. J. R., ``Modelling of non-linear active and passive
devices in three dimensional TLM networks,'' IEEE MTT-S Symp. Dig., 1997, pp. 383
386.
14. Ciampolini, P., Roselli, L., and Stopponi, G., ``Integrated FD-TD and solid state device
simulation,'' IEEE Microwave Guided Wave Lett., vol. 6, no. 11, Nov. 1996, pp. 419421.
15. Imtiaz, S. M. S., and El-Ghazaly, S. M., ``Global modeling of millimeter-wave circuits:
electromagnetic simulation of ampliers,'' IEEE Trans. Microwave Theory Tech. vol. 45,
no. 12, pt. 2, Dec. 1997, pp. 22082216.