Hartung-Gorre Verlag
Inh.: Dr.
Renate Gorre D-78465
Konstanz Fon: +49 (0)7533 97227 Fax: +49 (0)7533 97228 www.hartung-gorre.de
|
S
|
Scientific Reports on Micro and Nanosystems
edited by Christofer Hierold
Vol. 28
Verena Maiwald
A Microelectromechanical Switch
for Bandpass Vibration Detection.
1st Edition 2018. XVI, 164 pages. € 64,00.
ISBN 978-3-86628-614-6
Abstract
Wireless sensor networks continue developing towards
spatially distributed nodes and the integration of a larger number of sensors.
Efforts to reduce the resulting high data rates and power consumption include
sleep/wake-up modes and asynchronous communication protocols. Distributed
sensing and local data pre-processing, where only relevant information is
disseminated through the network is often seen as the way to proceed. On the
level of the sensor node the goal is autonomous operation powered by batteries
and energy harvesters. This development has implications for the sensors
integrated in such networks. They need to be low cost, low power and easy to
integrate with digital circuits. This consideration shifts the demand away from
high performance, high data rate, sensing that is local and of short duration
to low power, always-on sensor front-ends. Commercial capacitive
microelectromechanical system (MEMS) accelerometers offering these features are
readily available. However, they typically cover the frequency range below 1
kHz. Structural Health Monitoring could benefit greatly from MEMS sensors, but
higher bandwidths are required. There, commercial, sensitive piezoelectric
devices are bulky and high priced while MEMS accelerometers consume too much
power at low sensitivities.
To address this demand, the feasibility of a MEMS for low kHz vibration sensing is explored in this
thesis. The system should integrate the functionality of a bandpass filter,
signal amplifier and threshold detector. This approach reduces the need for
continuous sampling at high frequencies, thus enabling vibration detection at
minimal power consumption. Sensor output levels large enough to wake up a
micro-controller in sleep mode are targeted. To achieve this, the approach of a
two-stage transduction principle is followed: First, the vibration is
dynamically amplified in a defined frequency range. This is done mechanically
using a chain of coupled mass-spring-resonators. Thereby, amplified
displacements and frequency selectivity are achieved at zero power expense. For
the second stage, viz threshold detection, an
electromechanical, dynamic, pull-in switch is chosen. When the output of the
mechanical amplifier exceeds a threshold, an electrical contact is closed.
While consuming only leakage current in off-state, comparably high output
currents are achieved in on-state.
In this work, several steps have been taken to
translate this concept into reality: First, the transduction principle is
developed. This results in a lumped element model for a generic broadband
pull-in vibration trigger consisting of n-resonators. The model is then
translated into a planar design for fabrication in Silicon on Insulator (SOI)
based micromachining. Second, a mechanical amplifier based on six resonators
coupled in series is fabricated and characterized using a differential
capacitive readout. A bandpass response from 3 kHz-13 kHz is achieved at a
minimum and mean mechanical amplification of 16.4 dB and 23.5 dB, respectively.
Subsequently, two types of Silicon-to-Silicon pull-in vibration switches are
fabricated using the state of the art, near-vacuum epitaxial sealing (epi-seal) process developed in the
Stanford Microstructures & Sensors Lab. The first type of switch is based
on a single oscillating mass and the second on a dynamic amplifier with a chain
of three resonators. Both devices only require a static voltage of 5.3 V for
operation and are tested using a time domain approach. They are used to detect
more than 23’000 vibration events each. This is done at an on/off power ratio
of >1000 (11 nW in
off-state vs. 13.6 μW in
on-state). Further, the dependence of detectors’ output states (open and
closed) depending on the vibration’s amplitude and frequency is investigated.
The corresponding switching map of the dynamic pull-in is simulated and
measured for a parameter space in frequency and amplitude. Close to the
resonances, tongues of instability (trigger on) are observed, similar to
literature. The response of the one mass trigger at the primary resonance has a
minimum detected threshold of 7.7 
4.1 nm. A
broadband response with multiple overlapping tongues is achieved using the three-resonators device with a bandwidth of 6-18.7 kHz at
5.3 V. The minimum and broadband detection threshold is 1.1 nm 0.2 nm and 17.5
2.2 nm,
respectively. One major advantage of the operation in the strongly nonlinear
regime is the lowering of the broadband threshold (factor 6 compared to the
linear case in the three resonators scenario). However, this also introduces a
more complex switching behavior. Within the coarse, tongue-shaped envelope, a
fine pattern of alternating bands of stable and unstable parameters is
resolved. Similar patterns have been predicted numerically in literature for
the dynamic pull-in of MEMS resonators. Here, they are also experimentally
demonstrated.
Finally, predictions are made for the performance of
future device generations. As a next step, increasing the number of resonators
in the mechanical amplifier and scaling down the capacitive gap can reduce the
triggering thresholds at no additional power expense. We expect median
broadband thresholds of 56 pm for a device that can be fabricated with existing
processes. Medium term goals include an ultra-low power readout circuit, vacuum
packaging of the up-scaled designs, and the optimization of the acoustic
interface to the surface under test. In conclusion, the operation of a
microelectromechanical broadband pull-in vibration trigger was successfully
demonstrated, opening a path towards ultra-low power broadband vibration
detection. The device only requires a static voltage at low leakage for
operation and provides output levels high enough to wake up a microcontroller
in sleep mode.
Keywords: sleep/wake-up modes, asynchronous communication
protocols, distributed sensing, local data pre-processing, autonomous operation
powered by batteries and energy harvesters, MEMS for low kHz vibration sensing,
vibration detection at minimal power consumption.
Scientific Reports on Micro and Nanosystems
Direkt bestellen bei / to order directly from:
Hartung-Gorre
Verlag / D-78465 Konstanz / Germany
Telefon: +49 (0) 7533 97227
Telefax: +49 (0) 7533 97228
http://www.hartung-gorre.de
eMail: verlag@hartung-gorre.de