Submillimeter-Wave 3.3-bit RF MEMS Phase Shifter Integrated in Micromachined Waveguide

This paper presents a submillimeter-wave 500–550-GHz MEMS-reconfigurable phase shifter, which is based on loading a micromachined rectangular waveguide with 9 E-plane stubs. The phase shifter uses MEMS-reconfigurable surfaces to individually block/unblock the E-plane stubs from the micromachined waveguide. Each MEMS-reconfigurable surface is designed so that in the nonblocking state, it allows the electromagnetic wave to pass freely through it into the stub, while in the blocking state, it serves as the roof of the main waveguide and blocks the wave propagation into the stub. The phase-shifter design comprises three micromachined chips that are mounted in the H-plane cuts of the rectangular waveguide. Experimental results of the first device prototypes show that the microelectromechanical system (MEMS)-reconfigurable phase shifter has a linear phase shift of 20
$\mathbf {^{\circ }}$
in ten discrete steps (3.3 bits). The measured insertion loss is better than 3 dB, of which only 0.5–1.5 dB is attributed to the MEMS surfaces and switched stubs, and the measured return loss is better than 15 dB in the design frequency band of 500–550 GHz. It is also shown that the major part of the insertion loss is attributed to misalignment and assembly uncertainties of the micromachined chips and the waveguide flanges, shown by simulations and reproducibility measurements. The MEMS-reconfigurable phase shifter is also operated in an analog tuning mode for high phase resolution. Furthermore, a detailed study has been carried out identifying the reason for the discrepancy between the simulated (90
$\mathbf {^{\circ }}$
) and the measured (20
$\mathbf {^{\circ }}$
) phase shift. Comb-drive actuators with spring constant variations between 2.13 and 8.71 N/m are used in the phase shifter design. An actuation voltage of 21.94 V with a reproducibility better than
$\mathbf {\sigma }=$
0.0503 V is measured for the actuator design with a spring constant of 2.13 N/m. Reliability measurement on this actuator was performed in an uncontrolled laboratory environment and showed no deterioration in the functioning of the actuator observed over one hundred million cycles.