Page 111 - Antennas for Base Stations in Wireless Communications
P. 111

84      Chapter Two

                    In a dual-band array, the coupling between high- and low-band ele-
                  ment groups can give rise to high-band radiating currents flowing in
                  low-band elements, and this causes frequency-dependent ripples or
                  other deformities in the high-band patterns of the array. The amplitude
                  of low-band currents flowing in high-band elements is usually small
                  and is seldom troublesome. In an array in which inputs at both bands
                  are combined using a diplexer, excessive coupling at the element may
                  interact with the behavior of the diplexer to cause unwanted ripples in
                  the input VSWR of the high-band array.

                  2.3.12.2  Feed Networks and Arrays  Before building a prototype array,
                  ensure the feed network provides the expected complex currents at its
                  output ports and is well matched when all the output ports are correctly
                  terminated. Although the network may have been simulated before it
                  was constructed, coaxial feed systems are likely to show some differences
                  in the stray reactances associated with cable junctions. Inaccuracy in
                  modeling of interline coupling in microstrip networks may require some
                  impedance correction on a physical model, particularly if line spacings
                  are small to economize on the use of expensive laminate materials.
                    A simple method for the diagnosis of feed network problems is to
                  sample the outputs of different elements of an array using a balanced
                  loop. As the loop is moved to couple to successive elements, the mea-
                  sured ratio of the currents in the sampled elements should be constant
                  in amplitude, with a phase difference having the correct nominal value
                  at midband and a value at other frequencies proportional to f/f 0 . This
                  measurement is easy to carry out using a VNA, and the result can be
                  displayed in polar or Cartesian form. A corresponding measurement
                  can be carried out on the feed network itself, terminated with matched
                  loads, and the difference between the two results is a clear indication
                  of the effects of element mismatch (including mutual coupling). The
                  periodicity of ripples in the response give an immediate indication of
                  the distance separating the sources of interacting reflections, allow-
                  ing the matching to be improved to reduce the excursions from the
                  wanted values. The results become confusing if the ports of the VNA are
                  not themselves well-matched (as is common), so matched attenuators
                  should be inserted in series at the VNA ports.
                    The feed network of a RET antenna can be measured in this way to
                  provide confirmation that there are no unwanted interactions within the
                  network and that the output currents are correct for the required range
                  of tilt angles. As an extension of this technique, it is possible to construct
                  an anechoic “box,” lined with absorber, with arrangements for a probe
                  to be scanned along the array to measure the relative complex element
                  currents. These sampled currents can be used to compute the elevation
                  pattern of the array, often with surprising accuracy. Measurements of
   106   107   108   109   110   111   112   113   114   115   116