Page 110 - Antennas for Base Stations in Wireless Communications

Base Station Antennas for Mobile Radio Systems 83

2.3.12 Array Optimization
and Fault Diagnosis
Once a prototype is constructed, it is often found that its performance
does not fully meet expectations. A modern dual-polar multiband
antenna is a complex structure, and determining what has gone wrong
when performance expectations are not met is not easy. Finding the
source of the problems requires some clear thinking and careful experi-
mental techniques.

2.3.12.1 Single Radiating Elements and Element Groups Measurements
on a single element or element group will usually include the measure-
ment of azimuth patterns, input impedance, polarization, and XPI. The
most troublesome parameter to contain over a wide bandwidth is the
azimuth beamwidth, and this is best optimized by a mixture of computer
simulation and experiment. Simulation is good for rapid optimization
of simple mechanical parameters, for example, the spacing between
the radiator and the reflector or the reflector width and the depth of a
flange on the front edge of the reflector. When the simulation doesn’t
produce the desired result, it may be useful to carry out some practical
experiments to provide new ideas about what can be done and then to
optimize the idea using the simulator.
It is usually worth checking the magnitude of the mutual impedance
between adjacent array elements to make sure the driven impedance
of the elements will remain within expected bounds when the phases
and amplitudes of currents in neighboring elements are changed as a
result of beam-shaping or electrical tilt. In a RET antenna, the effect
of mutual impedance will be to modify the complex currents in the
array elements in unintended ways when the beamtilt changes, causing
unwanted variations in the elevation pattern. In an array covering a
single band group, mutual impedance can be controlled with separat-
ing fences between the elements, but this solution is not easy to apply
to a dual-band antenna where adjacent elements are of very differ-
ent dimensions. For this reason, the elevation pattern performance for
dual-band interleaved arrays tends to be less well controlled than for
separate low-band and high-band arrays.
A high value of measured XPI for a single tier does not ensure that the
performance of the complete array will be adequate, but if the isolation
of the components of a single tier is significantly less than the required
30 dB for the complete array, obtaining 30 dB from the complete array
will be difficult. It is always hard to compensate a local effect by a com-
pensating coupling at some point electrically far away; such compensa-
tion tends to work over some relatively narrow band where the phase
relationships between the error vector and the compensating coupling
are correctly phased.