27 January 2020 in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control
A 61-MHz Pierce oscillator constructed in 0.35μm CMOS technology and referenced to a polysilicon surface-mi-cromachined capacitive-gap-transduced wine glass disk resonator has achieved phase noise marks of -119 dBc/Hz at 1-kHz offset and -139 dBc/Hz at far-from-carrier offsets. When divided down to 13 MHz, this corresponds to -132 dBc/Hz at 1-kHz offset from the carrier and -152 dBc/Hz far-from-carrier, sufficient for mobile phone reference oscillator applications, using a single MEMS res-onator, i.e., without the need to array multiple resonators. Key to achieving these marks is a Pierce-based circuit design that har-nesses a MEMS-enabled input-to-output shunt capacitance more than 100× smaller than exhibited by macroscopic quartz crystals to enable enough negative resistance to instigate and sustain oscil-lation while consuming only 78 μW of power–a reduction of ~4.5× over previous work. Increasing the bias voltage of the resonator by 1.25V further reduces power consumption to 43μW at the cost of only a few dB in far-from-carrier phase noise. This oscillator achieves a 1-kHz-offset figure of merit (FoM) of -231dB, which is now the best among published chip-scale oscillators to date. A complete linear circuit analysis quantifies the influence of resona-tor input-to-output shunt capacitance on power consumption and predicts further reductions in power consumption via reduction of electrode-to-resonator transducer gaps and bond pad sizes. The demonstrated phase noise and power consumption posted by this tiny MEMS-based oscillator are attractive as potential enablers for low power “set-and-forget” autonomous sensor networks and embedded radios.
16 July 2014 in Journal of Microelectromechanical Systems
A capacitive-piezoelectric (also known as, capacitive-piezo) transducer that combines the strengths of capacitive and piezoelectric mechanisms to achieve a combination of electromechanical coupling and Q higher than otherwise attainable by either mechanism separately, has allowed demonstration of a 1.2-GHz contour-mode aluminum nitride (AlN) ring resonator with Q > 3000 on par with the highest measured d 31 -transduced AlN-only piezoelectric resonators past 1 GHz, and a 50-MHz disk array with an even higher Q > 12 000. Here, the key innovation is to separate the piezoelectric resonator from its metal electrodes by tiny gaps to eliminate metal material and metal-to-piezoelectric interface losses thought to limit thin-film piezoelectric resonator Q, while also maintaining high electric field strength to preserve a strong piezoelectric effect. While Q increases, electromechanical coupling decreases, but the k eff 2 · Q product can still increase overall. More importantly, use of the capacitive-piezo transducer allows a designer to trade electromechanical coupling for Q, providing a very useful method to tailor Q and coupling for narrowband radio frequency (RF) channel-selecting filters for which Q trumps coupling. This capacitive-piezo transducer concept does not require dc-bias voltages and allows for much thicker electrodes that reduce series resistance without mass loading the resonant structure. The latter is especially important as resonators and their supports continue to scale toward even higher frequencies. [2013-0395].
16 November 2018 in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control
This part one of two papers introduces a design flow for micromechanical RF channel-select filters with tiny fractional bandwidths capable of eliminating strong adjacent channel block-ers directly after the antenna, hence reducing the dynamic range requirement of subsequent stages in an RF front-end. Much like VLSI transistor circuit design, the mechanical circuit design flow described herein is hierarchical with a design stack built upon vi-brating micromechanical disk building blocks capable of Q’s ex-ceeding 10,000 that enable low filter passband loss for tiny frac-tional bandwidths. Array-composites of half-wavelength coupled identical vibrating disks constitute a second level of hierarchy that reduces the filter termination impedance. A next level of hierarchy couples array-composites with full-wavelength beams to affect fully-balanced differential operation. Finally, identical differential blocks coupled with quarter-wavelength beams generate the de-sired passband. Part II of this study corroborates the efficacy of this design hierarchy via experimental results that introduce a 39nm-gap capacitive transducer, voltage-controlled frequency tuning, and differential operation towards demonstration of a 0.1% bandwidth, 223.4-MHz channel-select filter with only 2.7dB of in-band insertion loss and 50dB of stopband rejection.
26 November 2018 in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control
This part II of a two-paper sequence presents fabri-cation and measurement results for a micromechanical disk-based RF channel-select filter designed using the theory and procedure of Part I. Successful demonstration of an actual filter required sev-eral practical additions to an ideal design, including the introduc-tion of a 39nm-gap capacitive transducer, voltage-controlled fre-quency tuning electrodes, and a stress relieving coupled array de-sign, all of which combine to enable a 0.1%-bandwidth 223.4-MHz channel-select filter with only 2.7dB of in-band insertion loss and 50dB rejection of out-of-band interferers. This amount of rejection is more than 23dB better than a previous capacitive-gap trans-duced filter design that did not benefit from sub-50nm gaps. It also comes in tandem with a 20dB shape factor of 2.7 realized by a hi-erarchical mechanical circuit design utilizing 206 micromechani-cal circuit elements, all contained in an area footprint of only 600μm×420μm. The key to such low insertion loss for this tiny per-cent bandwidth is Q’s >8,800 supplied by polysilicon disk resona-tors employing for the first time capacitive transducer gaps small enough to generate coupling strengths of Cx/Co ~0.1%, which is a 6.1× improvement over previous efforts. The filter structure uti-lizes electrical tuning to correct frequency mismatches due to pro-cess variations, where a dc tuning voltage of 12.1V improves the filter insertion loss by 1.8dB and yields the desired equiripple pass-band shape. Measured filter performance, both in- and out-of-channel, compare well with predictions of an electrical equivalent circuit that captures not only the ideal filter response, but also par-asitic non-idealities that distort somewhat the filter response.