Two separate teams of corporate researchers are engaged in programs to develop and demonstrate HTS technology for wireless applications in Japan. One team, AMTEL, has 21 researchers at work on system, filter, cryogenic refrigerator, and interface topics. The technical talent has been drawn from ALPS Electric and Denso, the two collaborating firms. Full details of the formation, funding, and business strategy for the venture are given in the AMTEL site report in Appendix B. A different group, known by the name "Western Alliance" is a collaboration between Matsushita, Sumitomo, and Kyocera. Unlike the AMTEL operation, which is principally located in a central site, the Western Alliance is conducted in a distributed fashion, with the microwave device work being the responsibility of the Matsushita researchers (see that site report in Appendix B). Sumitomo, which in the late 1980s had the leading program in Japan on thin-film microwave filters, withdrew from the device work in the early 1990s and is committed to the alliance only for film supply, not for device development. In addition to these two corporate teams, there is an ISTEC effort that is relatively broad and comprises not only the development of microwave filters, but the exploration of integrated mixer antenna structures at frequencies well above the commercial wireless bands.
The three groups developing microwave devices have internal team resources for production of HTS thin films on LaAlO3 and/or MgO substrates. As a representative example of the maturity of the device efforts in Japan, the 1996 filter results from AMTEL are reproduced in Fig. 4.2.
AMTEL is focused on bent (or hair-pin) microstrip resonator structures for creating receive filters. The 9- pole, 2.6 GHz filter in the above figure had a 34 MHz 3 dB bandwidth with 0.25 dB insertion loss. The 40 dB bandwidth was a relatively broad 68 MHz with a virtual zero in the low-frequency skirt and a slower than ideal roll-down in the high-frequency skirt. These distortions are indicative of the typical difficulty of implementing a filter response uncorrupted by cross-talk between nonadjacent resonators.
The most complex filter described by any group in Japan is the 9-pole result displayed in Fig. 4.2. This 9- pole response is a respectable initial demonstration and is correctly categorized as such by the AMTEL group as the end result of the 3-year "fundamental study" phase of its development of filters. Many additional details must be addressed during the scheduled 2-year phase (1997-98) in which a filter subsystem is to be created. Note that the measured response is shifted from the design response, and this problem will be exacerbated by the need to further reduce the fractional bandwidth so that the pass band is only 1-15 MHz wide, depending on the specific application. In-band ripples, too, must be reduced. Finally, more poles are required for a practical filter in order to achieve the sharp skirts required in the crowded wireless bands.
At Matsushita, within the Western Alliance, the early focus was on stripline structures, but that has shifted to disc-like planar configurations with the intention of reducing current crowding and thereby achieving higher power handling. Single resonator structures implemented in double-side-coated substrates have achieved Q values to 8,000, and based on measurements with up to 15 W of applied power, Matsushita researchers project a third-order intermodulation distortion (IMD) intercept of about 300 W for frequencies up to 5 GHz. These values fall short of results for similar structures reported in the international community. For example, the University of Wuppertal and Bosch reported at the ISEC 97 meeting in Berlin measurements on a 2-pole disc filter at 77 K with up to 100 W transmitted power. This was equivalent to 10 kW of circulating power, and the resonators degraded in unloaded Q from 50,000 to only 8,000 with a concomitant increase in insertion loss for the filter of 0.1 dB. Similar impressive results have been achieved by MIT Lincoln Lab and Lucent Technologies. In the area of power handling, then, the work in Japan is less mature. In addition, the extension of these disc resonator configurations to a complex multipole filter is extremely difficult to model, and the plan for such extension by Matsushita was not clearly described during the WTEC visit.
Among the Japanese researchers there appears to be a general optimism that standard modeling tools and precise manufacturing techniques will permit the reproducible fabrication of complex filters with accurate frequency response. This has not been the experience of U.S. researchers. Japanese optimism will likely dissipate rapidly as the next level of filter performance is attempted by the Japanese groups. This same optimism pervaded the U.S. activities about 2 years prior to this study; subsequent efforts to demonstrate practical subsystems for the wireless community resulted in the need to mechanically trim filter response, even after successive iterations of the design were done with empirical "tweaks" made to the computer algorithms in efforts to account for the special nature of HTS films and substrates. This difficulty is one of the motivations for a new DARPA-sponsored initiative: Frequency Agile Materials for Microelectronics.