Okay, here is the abstract and summary for the provided transcript on MEMS-based oscillators.
Abstract:
This tutorial, presented by Professor Clark Boyden from UC Berkeley, provides a comprehensive overview of MEMS-based oscillators. Starting with his background in academia, DARPA, and industry (founding Discera), the talk covers fundamental MEMS fabrication techniques (surface and bulk micromachining, epi-seal packaging), application drivers (communications, sensors, GPS, low-power systems), and basic oscillator theory (role of Q, phase noise, Leeson's equation). The core focuses on vibrating MEMS resonators, comparing capacitive and piezoelectric transduction. Various geometries like beams (clamped-clamped, free-free), disks, and rings are discussed, highlighting design techniques (nodal supports, quarter-wave stubs) to achieve high Q factors by minimizing anchor loss. A key emphasis is placed on capacitive resonators, detailing how scaling the electrode-resonator gap dramatically reduces motional resistance and increases coupling strength (KT^2), with experimental results shown down to 13nm gaps. The tutorial contrasts MEMS oscillator design (e.g., Pierce) with quartz, demonstrating the significant power consumption advantage of MEMS due to its inherently low parasitic capacitance (C0). Challenges like acceleration sensitivity (where MEMS excels), aging, and thermal drift are addressed, culminating in the concept of ultra-low power MEMS-based Oven-Controlled Oscillators (OCXOs) enabled by the superior thermal isolation achievable with micro-scale structures.
MEMS-Based Oscillators Tutorial Summary
0:00:07 Introduction: Professor Clark Boyden (UC Berkeley) introduces the tutorial topic: MEMS-based oscillators.
0:00:26 Speaker Background: Details experience at UC Berkeley, UMichigan, DARPA (funding MEMS research like CSAC), founding Discera (first MEMS oscillator company), and presidency of UFSC. Provides context for his expertise.
0:05:20 MEMS Fabrication - Surface Micromachining: Explains a common MEMS fabrication process using deposited layers (polysilicon structural layer, oxide sacrificial layer) and etching to create released, movable structures like beams.
0:09:20 MEMS Fabrication - Bulk Micromachining: Describes an alternative using the silicon wafer itself as the structural material, often involving deep etching (DRIE) and wafer bonding, allowing for thicker devices.
0:11:43 MEMS Packaging - Epi-Seal: Highlights the critical importance of packaging for stable oscillators. Details the Epi-Seal process (used by SiTime) which creates a very clean, sealed vacuum environment during fabrication, crucial for stability and Q.
0:15:13 MEMS and CMOS Integration: Discusses the potential and challenges of integrating MEMS directly with CMOS circuits. Key difficulty is thermal budget compatibility. Main advantage is minimizing parasitic capacitance.
0:17:36 Application Drivers for MEMS Oscillators: Better timing enables secure communications, efficient spectrum use, GPS improvements, lower power sensor networks, and superior resilience to jamming. MEMS offers potential advantages in power and size.
0:19:14 MEMS in Wireless Communications: Discusses replacing quartz crystals and potentially RF filters (like FBARs) in devices like cell phones, aiming towards single-chip radios.
0:20:45 Low-Power Applications (Super-Regenerative Receiver): Shows how low-power MEMS oscillators enable alternative, simple receiver architectures suitable for ultra-low power applications like sensor networks.
0:22:18 Oscillator Fundamentals - Role of Q: Explains oscillators as a tank circuit + amplifier in feedback. High Quality Factor (Q) is crucial for stability because it results in a steep phase slope near resonance, minimizing frequency shifts due to phase variations in the loop.
0:40:04 Oscillator Fundamentals - Phase Noise: Explains that noise (thermal, device noise) in the oscillator circuit creates phase/amplitude fluctuations, resulting in energy spreading into sidebands around the carrier frequency. High Q helps filter this noise, improving phase noise performance, which is critical for communication systems.
0:45:05 Oscillator Theory (Noise Filtering View): Presents oscillators as devices that amplify and extremely narrowly filter inherent noise sources (like thermal noise). Positive feedback dramatically increases the effective Q of the tank, making the output appear stable over long periods (long tau).
0:50:51 Phase Noise Equation (Leeson's Model): Shows that phase noise depends inversely on resonator Q squared and the power circulating in the oscillator loop, and directly on the noise factor (F) and temperature (kT).
0:54:10 Oscillator Types & PLLs: Contrasts high-Q/low-frequency (quartz) vs. low-Q/high-frequency (on-chip LC/ring) oscillators. Phase-Locked Loops (PLLs) are often used to combine the benefits, locking a low-phase-noise reference (like quartz or MEMS) to a higher frequency Voltage-Controlled Oscillator (VCO).
1:00:30 MEMS Resonator Basics (Clamped-Clamped Beam): Introduces vibrating mechanical structures as resonators. Frequency depends on stiffness/mass. Shows a polysilicon beam example (MHz range). Explains capacitive transduction (DC bias + AC voltage drive) and the concept of voltage tunability via electrical stiffness. Takeaway: Requires vacuum for high Q.
01:09:12 High-Frequency Limitation - Anchor Loss: Scaling simple clamped-clamped beams to higher frequencies leads to significant Q degradation due to energy leaking through the anchor points into the substrate.
01:11:15 Improved Design - Free-Free Beams: Introduces a superior design using a "free-free" beam supported at nodal points by quarter-wavelength structures. This isolates the resonator from the anchors, dramatically reducing anchor loss and restoring high Q at higher frequencies (e.g., Q=15k @ 92 MHz).
01:17:53 Higher Frequencies - Disk Resonators: Introduces disk-shaped resonators operating in radial contour modes, anchored centrally for low loss. Frequency is set by lithographically defined radius. Q > 10k demonstrated at 1.5 GHz. Key challenge: High motional resistance (Rx).
01:22:31 Motional Resistance & Gap Scaling: Explains that motional resistance (Rx) for capacitive resonators is strongly dependent on the electrode-resonator gap (d), scaling as d^4. Takeaway: Reducing the gap is extremely effective for lowering Rx and improving electromechanical coupling.
01:29:17 Fabrication for Nano-Gaps: Details the process using sacrificial sidewall spacers to define gaps down to the nanometer scale (e.g., 80nm, 40nm, 13nm), requiring smooth etched sidewalls. Polydiamond is highlighted as a high-Q material.
01:37:04 High-Q Ring Resonators: Shows spoke-supported ring resonators achieving exceptional Q factors (77k @ 900 MHz, 42k @ 3 GHz) in diamond. Problem remains the high Rx with the demonstrated ~75nm gaps.
01:40:55 Transducer Comparison: Capacitive vs. Piezo: Capacitive offers potentially higher Q but weaker coupling (unless gap is tiny). Piezoelectric offers stronger coupling (lower Rx) but typically lower Q. Argues that improving capacitive coupling (by shrinking gap) is more promising than improving piezo Q.
01:43:15 Impact of Gap Scaling: Simulation shows shrinking the gap from 80nm towards 10nm drastically reduces Rx and increases coupling factor (KT^2) into percentage range, making capacitive transduction very competitive.
01:46:02 Experimental 13nm Gaps: Presents measured results for a 60 MHz wineglass disk resonator with a 13.2 nm gap, achieving low motional resistance (< 100 Ohms) with low
Error: int() argument must be a string, a bytes-like object or a real number, not 'NoneType' voltage (5.5V), confirming the effectiveness of gap scaling.
01:51:38 MEMS Power Advantage - Oscillator Design: Compares Pierce oscillator design using quartz vs. MEMS. The crucial difference is the resonator's shunt parasitic capacitance (C0), which is pico-Farads for quartz but femto-Farads for MEMS. Lower C0 allows much smaller loop capacitors (C1, C2), requiring significantly less drive current and thus lower power consumption for the same performance. Takeaway: Low C0 is a fundamental MEMS advantage for low-power oscillators.
01:59:47 Measured Low-Power, High-Performance Oscillator: Demonstrates a 60 MHz MEMS oscillator meeting GSM phase noise requirements while consuming only 78-110 microwatts, orders of magnitude lower than typical quartz solutions.
02:01:23 Improving Performance - Resonator Arrays: Shows that mechanically coupling multiple resonators in an array increases power handling and reduces overall motional resistance (Rx goes down by factor of N), leading to improved phase noise.
02:05:04 Acceleration Sensitivity: MEMS resonators are inherently less sensitive to acceleration and vibration than quartz due to their small mass and high stiffness. Measured results show excellent performance (~1e-10 /g).
02:07:34 Aging and Long-Term Drift: Preliminary aging tests on a poly-Si MEMS resonator with oven control show good stability (<50 ppb) after initial burn-in, indicating promising long-term performance.
02:09:40 Thermal Drift & Ultra-Low Power OCXOs: The best stability comes from eliminating temperature changes (oven control). MEMS structures have vastly higher thermal resistance and lower thermal capacitance than quartz assemblies. Takeaway: This allows MEMS OCXOs to operate with milli-watts of heating power compared to watts for quartz OCXOs, enabling miniature, high-stability, low-power timing devices.