While the International Microwave Symposium is mainly a gathering where the latest microwave technology is displayed to the entire industry, on this occasion the past and future of microwave technology were also represented.  They serve as a reminder that while this year’s IMS 2006 was impressive, it represents a mere page in the continuing saga of microwave technology.

In one corner of San Francisco’s Mascone South convention center, where this year’s exhibit took place, there was a historical exhibit that included books, images and real components.  One component was a 1941 audio oscillator from HP.  Another component displayed nearby would have tremendous historical implications shortly after it was invented in 1937.  This was the klystron.

The chairman for this historical exhibit, a retired microwave engineer, related his opinion that the Varian Brothers played a major, though subtle part in history during the years immediately preceding World War II with their role in inventing the klystron. These men apparently saw a storm brewing and viewed the development of the klystron as an appropriate response, realizing its wartime potential. Their efforts had roots in Stanford physicist W.W. Hanson’s work on microwave resonators.

The company Varian Associates was established to develop the klystron, and manufacturing was later transferred to Sperry Corporation. Eventually, Varian got out of the vacuum tube business altogether.

While various tube technologies, including the traveling wave tube amplifier that was developed for wide bandwidth, in contrast to the klystron’s high-Q, fixed-frequency operation, can be readily distinguished from each other, the MMICs at the historical exhibit were much harder to differentiate for obvious reasons. The larger geometries of the earlier chips, along with their larger dye sizes, were the only visual indications of the nearly 40 years that span from Texas Instruments’ 1967 Silicon PIN Switch, perhaps the first MMIC, to Varian’s 1990 monolithic 5 GHz to 100 GHz InP Distributed Amplifier, and on to the latest MMICs on exhibition at IMS 2006.

One MMIC on exhibition was TLC Precision Wafer Technologies’ Linear Phase Shifter. According to Dr. Timothy Childs, President of TLC, the highly linear relationship between the phase shift of the input signal and the applied dc control voltage is achieved with precise geometric wafer structures and design configuration developed on a Cray supercomputer with the aid of NASA. These MMICs operate from the X-band through the W-band, and have a control voltage range from 0 V to 12 V, allowing continous or digitized phase-shifting/modulation (2 bits to 2000 bits) to 360 °.

One area of semiconductors that is quite different from dye-based RFICs is wafer-level technology. In a private discussion at IMS 2006, Dr. Julio A Navarro, a designer of active millimeter-wave phased-array antenna systems at Boeing-Phantom Works, related that hybrid techniques prove useful for solving technical challenges in wafer product development.

For example, the active elements may be implemented on one wafer, and the geometric antenna structures may be implemented on another wafer, while precision passive support components could be placed on yet another wafer. The wafers would then be stacked to form a complete system. Visual alignment, pick-and-place tolerances, compliant/reliable/thermally-electrically conductive die-attach materials and reliable/compliant vertical interconnects currently limit the operating frequency and complexity of such integrations.

However, just as with dye-based semiconductor technologies, where system-in-package techniques generally give way to system-on-chip techniques, it may become routine to implement certain types of systems, such as phased-array radars, monolithically on a single wafer. Some of the issues associated with complete heterogeneous wafer-level integration are detailed in Dr. Navarro’s book, “Integrated Active Antennas and Spatial Power Combining”.

One technological development that will advance the development of these systems is the capability to define multiple active device layers using silicon-on-insulator (SOI) technology. Other advances could come from competing micro- and nano-scale technologies. However, the initial power levels of wafer-level systems will be low, and yield and thermal management issues will introduce new sets of design challenges.

Dr. Navarro stated there are many competing requirements when active devices are combined with passive devices and components having transmission lines with antennas. Each has its own optimized variables, and it is uncertain whether or not combining everything reduces the benefits of individually optimizing each part. In a recent publication delivered to the GaAs IC symposium, Dr. Navarro did not recommend complete wafer-scale integration.

While semiconductor technology dominates the RF industry at present, mechanical radio technology continues to evolve. This is readily apparent in the resonant MEMS structures presented at IMS 2006, especially when compared to the mechanical coherer technology used to detect signals in the earliest radio systems in the late 1800s.

MEMS technology is well established in Tx/Rx antenna switching applications for RF signals. However, the switching action of these devices is much slower than the frequency of the conducted RF signals passing through them. In contrast, the mechanical vibrations of resonant MEMS structures can reach several Giga-hertz. Furthermore, these structures can operate with Q factors exceeding 10,000, making them superior to quartz crystals (which also utilize mechanical vibration to achieve high precision). While this allows these MEMS structures to function as narrowband filters and oscillators, yield, speed and reliability are still issues being addressed.

An introduction to MEMS resonators and filters was provided as a tutorial session at IMS 2006. One presentation presented individual MEMS components assembled into mechanical circuits that can be analyzed and optimized in the same manner as electronic circuits. This process builds on the mathematical similarities between electrical and mechanical systems, such as the similarity of a spring-mass system to an L-C tank circuit.

Recently developed MEMS structures and their associated technical challenges were also covered. For example, one MEMS structure, called a micromechanical guitar string for its longitudinal vibration mode, has a Q factor of 8000 in a vacuum that drops to about 50 in air. Another structure, the diamond disk, is essentially immune to the effects of air due to its uniform radial mode of oscillation, expanding and contracting within the plane of the disk about its center. This structure can have a Q factor of 10,000 in air and 11,500 in a vacuum. While maintaining a vacuum presents packaging challenges for dynamic MEMS structures, it also enables a new class of vacuum microelectronics based on static MEMS structures.

Just as vacuum tubes replaced electromechanical relays for switching applications, vacuum electronics implemented with MEMS technology is an emerging frontier of research. Specifically, pointed diamond structures can emit electrons in the same manner as the cathode of a vacuum tube, and a working triode based on this technology has been fabricated at Vanderbilt University. Vacuum microelectronics can provide the functionality of vacuum tubes without the production of heat, and may overcome high voltage and high power limitations in semiconductor technology. Though this topic was not covered at IMS 2006, it relevance to microwave technology will hopefully be discussed and demonstrated at future sessions.

For now, vacuum tubes will remain the exclusive technology where high RF transmitter power is required. Harry Rutstein of Dorado International, an exhibitor at IMS 2006 specializing in microwave/millimeter wave products, stated that at present there is no substitute for traveling wave tubes in high power RF amplifier applications. In the early 1950’s Rutstein’s company Eastern Associates once represented the Huggins Company, which made the HA1, which operated from 1 GHz to 2 GHz, and the 2 GHz to 4 GHz HA2. These tubes were the first commercial traveling wave tubes and were made with glass. They provided 10 W to 20 W of output power. Later on, other companies would develop ceramic tubes to generate greater power levels.

Another IMS exhibitor, Pendulum Electromagnetics, makes a line of 2 kW CW RF PAs based on traveling wave tube technology that operate from 0.4 GHz to 8 GHz in octave-band increments at 25% efficiency. Representatives from Pendulum, Carl and Mike Everleigh, and John Gatsis, noted that while semiconductor technology is beginning to penetrate into L-Band and S-Band applications, high power pulsed and continuous wave RF applications are still dominated by tube technology. However, engineering challenges remain in the areas of switching power supply ripple, linearity, reliability and cost, especially for communications-grade traveling-wave tube amplifiers (TWTAs).

The convergence of semiconductor electronics, dynamic MEMS structures, and vacuum microelectronics in the distant future is an intriguing possibility. These potential advances, taken together with the technology exhibited at IMS 2006, call into question whether any RF microwave technology can ever truly become obsolete.