SST Wireless
The AUTO21 Network of Centres of Excellence Inertial Sensor Cluster for Adaptive Path Prediction (F304-FIS)
Edmond Cretu (UBC) Bill Morton (SST Wireless) Ottawa May 2011
Some facts •
Motion sensing market: 752M units of MEMS accelerometers and gyroscopes produced worldwide in 2008
•
Automotive applications still dominant (increasing consumer applications share)
•
Trends: •
•
•
Global 2007-2013 market for MEMS accelerometers and gyroscopes (source: Yole Dev.)
Add intelligence as product differentiator (integration with signal processing)
•
Emergence of low-cost sensor clusters (e.g. IMU)
•
Technology advances in fabrication and packaging
Inertial Measurement Units: market of $1.55B in 2009, with a 9% annual growth rate From “classic” applications (accelerometers as crash sensors, gyroscopes for roll-over protection) to integrated modules (inertial navigation)
F304-FIS: Inertial Sensor Cluster for Adaptive Path Prediction Goal: Inertial Measurement Unit with 3DOF (ax,ay,Ωz)
Team: • Edmond Cretu (UBC) – MEMS and microsystems design • Shahriar Mirabbasi (UBC) – analog CMOS interface • Sima Mihai (U Vic) – FPGA system-level integration • Walied Moussa (U Alberta) – assembly and packaging • HQP team of 8 graduate students Supporting companies: SST Wireless (BC), Micralyne (AB), CMC Microsystems
Automotive target applications (1) •
Adaptive forward lightning (AFL) system: the orientation of the front headlamps is adjusted, depending on speed and curve severity (estimated by IMU), to provide better visibility with respect to the steering direction.
•
Improved active safety system: safer night-time and sinuous roads driving
•
Need: more than 40% of all automobile accidents resulting in death occur at night, when the traffic is 80% less! (German Federal Statistics Bureau)
•
IMU estimates the driving path, and provides control signals to a rotating system (~15deg) for the front headlamps.
• •
Concept introduced by Vauxhall Motors and Opel in 2008 (Vauxhall Insignia) 2010: Buick La Crosse
Automotive target applications (2) •
Path prediction for inertial navigation – combined local IMU and GPS navigation
•
Need: a new generation of digital road mapping – lane-detection and lane-change information incorporated into the navigation guidance systems •
•
2009: EGNOS (the European Geostationary Navigation Overlay Service) - new GPS network within EU, with an improved accuracy (2m) compared to conventional GPS (20m)
Other active safety systems: path prediction for intelligent collision avoidance systems -> the trajectory of the car is estimated versus the obstacles detected using other sensing systems (radar), and warning signals provided to the driver.
Collaborators •
General support: SST Wireless, CMC Microsystems, Micralyne, Melexis Belgium
•
Roles: •
Involved in defining the specifications needed for an general inertial measurement unit to be used in advanced automotive applications (SST Wireless, Melexis)
•
Estimate market, define specific requirements (e.g. self-calibration)
•
Support for software design and simulation tools (CMC Microsystems), fabrication and packaging services (CMC Microsystems and Micralyne)
Provide a test and measurement environment (SST Wireless)
Wireless interface with central control module (SST Wireless)
Accuracy you can count on! TPMS… Tire Pressure Monitoring Systems
SST Wireless - Technology Wireless Expertise
RF, microwave and millimeter wave
Bi-directional transceiver, PLL synthesizer, Low Power Receiver
•
A Leader in Integrated Circuits and Micro Systems – – – –
•
System on Chip, Lab on Chip Intelligent Micro-Nano-Technology Sensors Low Power Technology Miniature and Light Weight Packaging
Applications: Network Software, Firmware & Middleware
True Tire Technology (TTT)
A Real Time Tire Heat and Pressure Monitoring System OTR JV Partner –
Why Utilize TPMS for Industrial and Commercial applications? •
Enhances fleet control of tire performance
•
Greatly enhances safety – reduces heat related delambs, roll-offs and tire fires
•
Reduces downtime & service expense
•
Improves and extends tread life (longevity)
•
Improves fuel economy
•
...and now in the case of the “Mining Sector” it has become a tool for increasing production
How does the SST Wireless TPMS system do this? •
By utilizing the latest “RFID Beacon” technology – transceiver based chip sets (meaning) the sensor is bi-directional, it both transmits and receives information
•
Currently the standard for industry has been a directional or transmit only chip (thus greatly reducing the systems adaptability and flexibility to the customer needs)
•
• • • •
•
Tire Sensor – “RFID BEACON” Monitors Tire Pressure, Temperature and Mileage • In Real Time Is a Permanent Rim or Side wall mount – Inside Tire The Transceiver based Technology allows for: - Resetting Parameters (Options) - Is a Digital Wireless Air Gauge (No yard checks required) Minimal 5 Year Life on Sensor Electronics
Currently We Monitor Three Key Variables •
Real Pressure – Accuracy within 1 PSI
•
Real Temperature – Accuracy within 1 degree Celsius
•
Mileage – Individual Tire (GPS Required)
• •
…the SST TPMS Sensor unit has proven to be the most accurate and robust sensor on the market
TPMS - ROI •
Customers value the environmental and economic benefits of TPMS but the safety benefits drive their demand…(Commercial)
•
Key value proposition: reduced fuel consumption, reduced annual tire cost, reduced downtime and road service expense…TPMS has now become a production tool…(OTR, Mining)
SST TPMS Product Categories OTR (Mining) Closed System – Harsh Environment Public Transit (Bus) Interface to Network – Low Profile Municipal and Commercial Waste Haulers GPS Integration - Driver Alert capabilities
TPMS – OTR and Commercial “Real World” •
Our Lab - Highland Valley Copper Kamloops, BC
•
Our own Ex Whatcom County 44 passenger Orion Mark 5 transit bus
A few of our Customers: - NACG (North American Construction Group) - Syncrude (Alberta, Tar Sands) - Copper Mountain Mines Inc, - Klempke (Alberta, Tar Sands) - King County Transit - Edmonton Transit - Edmonton Waste - Vancouver Waste
General IMU microsystem design strategy •
Present generation of automotive MEMS sensors – suitable for air bag systems, vehicle stability systems, etc., but not enough performance/low cost ratio for inertial navigation
•
Estimated resolution (input from Melexis and SST Wireless):
•
•
Linear accelerations (ax, ay): 0.001g in 40Hz bandwidth
•
Angular rate (Ωz) sensing: 0.01deg/sec in 30Hz bandwidth
Design strategy: operate the microdevices in a strong (nonlinear) electromechanical coupling regime -> advanced digital signal processing algorithms in FPGA, using analog CMOS as a thin interface.
System level view
Global challenges and ways of managing them •
Complex microsystem: combination of MEMS devices (accelerometers and gyroscopes), CMOS analog electronics and digital signal processing implemented in reprogrammable hardware
•
Different design tools and simulation levels: finite elements (for MEMS and packaging), Spice, VHDL/Verilog, Simulink/Matlab
•
How do they fit together? •
Extract reduced order behavioral models in VHDL-AMS/Verilog-A for the inertial MEMS devices, to be used in the design and simulation of the analog CMOS subsystem
•
Simplified behavioral models (from simulations or experimental measurements) for the subsystems, to use them in Simulink system level architecture
•
Make possible (through design) the independent test of the subsystems
Inertial MEMS sensors •
Challenges: higher resolution and accuracy in a small die area, taking into account environment (e.g. temperature) fluctuations
•
The intrinsic mechanical sensitivity ~ inertial mass => bad scaling!!
•
Exploit innovative nonlinear coupling effects: operation on the stability border for accelerometers, parametric amplification for gyroscopes
•
Closed loop configuration: thin analog layer (capacitive sensing and actuation), with most of the feedback control in the digital domain
•
Silicon-on-insulator fabrication technologies: MEMSCap SOI-MUMPs, Tronics (fabrication services subsidized by CMC Microsystems)
High-sensitivity accelerometers (1) •
Pull-in based MEMS accelerometers – operation on the stability border, in order to get micro-g resolution
•
Collaboration with Delft University of Technology, The Netherlands, and University of Porto, Portugal
•
Fabrication technology: SOI-MUMPs (MEMSCAP)
Operating principle
•
Exploit the high sensitivity of the pull-in time to external forces in electrostatically operated MEMS devices -> resolution set primarily by the resolution of the time measurement
Pull-in accelerometer performance
a=
Vstep V pi
•
Calibrated measurements (using a reference optical fiber Bragg accelerometer) have shown a resolution of 0.25us/ug, in a measurement noise floor of about 400ug (intrinsic mechano-thermal noise ~ 40ug for a bandwidth~2/tpi = 180Hz)
•
Simple electronics for a->time conversion, but nonlinear dependence => conversion table or digital interpolation necessary
•
Challenge: the damping coefficient (and implicitly tpi ) strongly depends on the operating temperature => self-calibration needed in operation!
Publications – pull-in accelerometer 1.
R.A. Dias, L. Mol, R.F. Wolffenbuttel, E. Cretu, L.A. Rocha [2011]- “Design of a time-based micro-g accelerometer,” in IEEE Sensors, in press
2.
R.A. Dias, E. Cretu, R.F. Wolffenbuttel, L.A. Rocha [2011]”Pull-in based ugresolution accelerometer: characterization and noise analysis,” Sensors and Actuators A: Physical, Feb 2011
3.
L.A. Rocha, R.A. Dias, E. Cretu, L. Mol, R.F. Wolffenbuttel [2011]”Auto-calibration of capacitive MEMS accelerometers based on pull-in voltage,” in Microsystem technologies, vol. 17, no. 3, pp. 429-436, Feb 2011
4.
R. Dias, E. Cretu, R. F. Wolffenbuttel, and L. Rocha [2010] “Characterization of a Pull-In Based µg resolution Accelerometer,” in Eurosensors XXIV
5.
R. A. Dias, R. F. Wolffenbuttel, E. Cretu, and L. A. Rocha, “Squeeze-film damper design with air channels: experimental verification,” in Eurosensors XXV, Athens, Greece, 2011, Sep 4-7
High-sensitivity accelerometers (2) – Closed-loop case •
Goal: digital control loop, easy to implement in FPGA
•
Classic control methodology: sigma-delta
•
Novel MEMS control architecture: sliding-mode control – two-level actuation
From simulation to experiment and back •
Co-simulation with analog electronics and Simulink/FPGA ->needs behavioral (reduced order models) Polytec MSA-500 micromotion analyzer
Digital binary feedback accelerometer
Self-calibration principle for inertial MEMS sensors Principle: use binary pseudorandom sequences - electrostatic actuation with small amplitude PN to recover the impulse response Self-test and self-calibration – requirements from SST Wireless LTI system
Reference model self-calibration architecture Tuning Block (Fel)
MEMS Device DUT
Read-out
FPGA
DAC PN Generator
MEMS DUT-R
DAC
ADC
-
+
Vout Calibration
Adaptive self-calibration
Self-calibration FPGA blocks PN Sequence generator
DUT-R
Experimental tests
A. Kansal, E. H. Sarraf, M. Sharma, and E. Cretu [2011] “Novel Adaptive FPGA-based selfcalibration and self-testing scheme with PN sequences for MEMS-based inertial sensors,” in IMS3TW 2011, Santa Barbara, California, USA, May 16-18, 2011
MEMS gyroscope
Nonlinear amplification techniques in MEMS gyroscopes •
Electro-mechanical parametric amplification for vibratory angular rate sensors: modulate electrically the stiffness of the sense mode
MEMS gyroscope publications 1.
M. Sharma, E. H. Sarraf, and E. Cretu [2011] “Parametric amplification/damping in MEMS gyroscopes,” in IEEE MEMS 2011, The 24th IEEE Int. Conf. on Micro Electro Mechanical Systems, Jan 23-27, 2011, Cancun, Mexico.
2.
I. Sabageh, V. Rajaraman, E. Cretu, and P. J. French [2010] “Design and Modelling of a Decoupled and Tunable (40–330 Hz) SOI-MEMS Yaw Rate Gyroscope,” in STW SAFE 2010 - Semiconductor Advances for future electronics and sensors, 2010.
3.
I. Sabageh, V. Rajaraman, E. Cretu, and P. J. French [2010] “Design and modeling of a three-mass, decoupled, tunable SOI-MEMS gyroscope with sense frame architecture,” in 21st Micromechanics and Microsystems Europe Workshop (MME2010), 2010.
4.
M. Sharma, E. H. Sarraf, and E. Cretu, “A novel dynamic pull in MEMS gyroscope,” in Eurosensors XXV, Athens, Greece, 2011, Sep 4-7
Capacitive sensing CMOS circuit
Low-noise parasitic-insensitive switched-capacitor-based CMOS front-end Output voltage is proportional to input capacitance variation The output voltage is digitized by an on-chip analog-to-digital converter
Analog CMOS interface – learned lessons •
The interface between MEMS and CMOS dies is critical – do not underestimate the parasitic effects in the real world
•
Design for test and calibration! – the CMOS chip should be testable in the absence of any MEMS die.
•
J. Shiah, H. Rashtian, and S. Mirabbasi, " A Low-Noise Parasitic-Insensitive CMOS Switched-Capacitor Interface Circuit for MEMS Capacitive Sensors," to be presented at IEEE International NEWCAS Conference, June 2011.
•
Alternative solution for incipient tests with FPGA: lower-performance PCB capacitive readout
Kalman filtering for path prediction •
We use the square-root Kalman filter -> needs less numerical precision for the same performance
•
Kalman filter is computationally demanding requiring: •
QR Decomposition
•
Matrix inversion
System level: Automotive Reconfigurable Guidance System
Analog Part Accelerometer
Kalman Filter QR Decomposition with CORDIC
Instrumentation Amplifier
Matrix Inversion
A/D Converter
Path estimation
Accelerometer (UBC) + Instrumentation Amplifier (UVic + UBC) Kalman Filter (QR Decomposition, Matrix Inversion) on FPGA (UVic) Path estimation (UVic + UBC) + A/D Converter (UBC)
QR Decomposition- Accuracy in FPGA •
Matrix inversion (required by the Kalman tracking block) is expensive for realtime computation => made using QR algorithm, implemented through CORDIC schemes in FPGA
•
Number of test vectors = 105
•
Accuracy threshold = 10-3
•
Number of CORDIC iterations = 18
Custom- designed reconfigurable hardware
Commercial (fine-grained) FPGAs:
Large penalties (silicon area, power consumption)
Good for prototyping and low-batch implementation
Novelty: Coarse-grained FPGA: ShEERA (Shift-enabled Embedded Reconfigurable Array)
Based on shift-enabled interconnection network
Good for vector rotations, QR decomposition, etc.
Generic architecture for an Automotive Reconfigurable Computing Platform (AReComP)
Shift-Enabled Embedded Reconfigurable Array
Assembly and packaging – U. Alberta Flip chip Facility Specification • Three technique for bonding • Applied heat on the placing tool for the device and highly controlled bottom heater for the substrate • Friction welding using Ultrasonic module • Force and epoxy bonding for cold bonding • Placement Accuracy: within 5 µm in tolerance
Ultrasonic module thermo module
• Max. Board Size:410 mm x 234 mm with potential for upgrades • Chip Sizes: 0.2 mm up to 85 mm with upgrade options depending on the bonding technique • Bonding Force: min. 0.1 N max. 500 N • Total Magnification: up to 230 x
Dispensing module
Force module
Integration of MEMS Sensors in Flexible Materials Integration of MEMS Sensors in Flexible Materials Integration techniques investigated • Solder ball pickup and placement tool design: This is done using a homemade tool designed to pick different size of balls and micro assemble them on the substrate using the same level of tolerance the system can offer for the devices. • Gold ball bumps placement using wire bonding. Characterization of the bonding strength with variations in the bonding parameters. • Flip chip assembly on flexible substrates using ultrasonic frication. Side effect issues such as damping of the ultrasonic vibration is investigated using bump stacking. • Conductive epoxy and anisotropic conductive epoxy will be tested for flexible PCBs • Potential application • Gharib, H.H., Mohamed M. EL Gowini, Moussa, W.A., "Developing a MEMSBased Smart Tire", Auto21, Windsor (2010). [2nd place/78 participating poster] F304-FIS
Conclusions •
Large, complex project – it is essential to ensure by design a good interface between the subsystems and their independent testability
•
The same module will present different views/models, depending on the design level: layout, finite element model, behavioral model in Spice/VHDL-AMS or Simulink, etc.
•
Several innovative concepts: nonlinear amplification techniques for the MEMS devices, parallel CORDIC architecture for fast matrix inversion in FPGA and dedicated, low-cost, coarse-grained reconfigurable die (SHEERA), novel switched-capacitor interface, novel packaging techniques
•
Challenging but rewarding work – 8 HQPs, large publication number
•
The on-going collaboration with SST Wireless is essential, in order to get a realistic feedback “from the trenches” and for the wireless interface with other automotive modules.
Edmond Cretu (UBC) Bill Morton (SST Wireless) Ottawa May 2011
Some facts •
Motion sensing market: 752M units of MEMS accelerometers and gyroscopes produced worldwide in 2008
•
Automotive applications still dominant (increasing consumer applications share)
•
Trends: •
•
•
Global 2007-2013 market for MEMS accelerometers and gyroscopes (source: Yole Dev.)
Add intelligence as product differentiator (integration with signal processing)
•
Emergence of low-cost sensor clusters (e.g. IMU)
•
Technology advances in fabrication and packaging
Inertial Measurement Units: market of $1.55B in 2009, with a 9% annual growth rate From “classic” applications (accelerometers as crash sensors, gyroscopes for roll-over protection) to integrated modules (inertial navigation)
F304-FIS: Inertial Sensor Cluster for Adaptive Path Prediction Goal: Inertial Measurement Unit with 3DOF (ax,ay,Ωz)
Team: • Edmond Cretu (UBC) – MEMS and microsystems design • Shahriar Mirabbasi (UBC) – analog CMOS interface • Sima Mihai (U Vic) – FPGA system-level integration • Walied Moussa (U Alberta) – assembly and packaging • HQP team of 8 graduate students Supporting companies: SST Wireless (BC), Micralyne (AB), CMC Microsystems
Automotive target applications (1) •
Adaptive forward lightning (AFL) system: the orientation of the front headlamps is adjusted, depending on speed and curve severity (estimated by IMU), to provide better visibility with respect to the steering direction.
•
Improved active safety system: safer night-time and sinuous roads driving
•
Need: more than 40% of all automobile accidents resulting in death occur at night, when the traffic is 80% less! (German Federal Statistics Bureau)
•
IMU estimates the driving path, and provides control signals to a rotating system (~15deg) for the front headlamps.
• •
Concept introduced by Vauxhall Motors and Opel in 2008 (Vauxhall Insignia) 2010: Buick La Crosse
Automotive target applications (2) •
Path prediction for inertial navigation – combined local IMU and GPS navigation
•
Need: a new generation of digital road mapping – lane-detection and lane-change information incorporated into the navigation guidance systems •
•
2009: EGNOS (the European Geostationary Navigation Overlay Service) - new GPS network within EU, with an improved accuracy (2m) compared to conventional GPS (20m)
Other active safety systems: path prediction for intelligent collision avoidance systems -> the trajectory of the car is estimated versus the obstacles detected using other sensing systems (radar), and warning signals provided to the driver.
Collaborators •
General support: SST Wireless, CMC Microsystems, Micralyne, Melexis Belgium
•
Roles: •
Involved in defining the specifications needed for an general inertial measurement unit to be used in advanced automotive applications (SST Wireless, Melexis)
•
Estimate market, define specific requirements (e.g. self-calibration)
•
Support for software design and simulation tools (CMC Microsystems), fabrication and packaging services (CMC Microsystems and Micralyne)
Provide a test and measurement environment (SST Wireless)
Wireless interface with central control module (SST Wireless)
Accuracy you can count on! TPMS… Tire Pressure Monitoring Systems
SST Wireless - Technology Wireless Expertise
RF, microwave and millimeter wave
Bi-directional transceiver, PLL synthesizer, Low Power Receiver
•
A Leader in Integrated Circuits and Micro Systems – – – –
•
System on Chip, Lab on Chip Intelligent Micro-Nano-Technology Sensors Low Power Technology Miniature and Light Weight Packaging
Applications: Network Software, Firmware & Middleware
True Tire Technology (TTT)
A Real Time Tire Heat and Pressure Monitoring System OTR JV Partner –
Why Utilize TPMS for Industrial and Commercial applications? •
Enhances fleet control of tire performance
•
Greatly enhances safety – reduces heat related delambs, roll-offs and tire fires
•
Reduces downtime & service expense
•
Improves and extends tread life (longevity)
•
Improves fuel economy
•
...and now in the case of the “Mining Sector” it has become a tool for increasing production
How does the SST Wireless TPMS system do this? •
By utilizing the latest “RFID Beacon” technology – transceiver based chip sets (meaning) the sensor is bi-directional, it both transmits and receives information
•
Currently the standard for industry has been a directional or transmit only chip (thus greatly reducing the systems adaptability and flexibility to the customer needs)
•
• • • •
•
Tire Sensor – “RFID BEACON” Monitors Tire Pressure, Temperature and Mileage • In Real Time Is a Permanent Rim or Side wall mount – Inside Tire The Transceiver based Technology allows for: - Resetting Parameters (Options) - Is a Digital Wireless Air Gauge (No yard checks required) Minimal 5 Year Life on Sensor Electronics
Currently We Monitor Three Key Variables •
Real Pressure – Accuracy within 1 PSI
•
Real Temperature – Accuracy within 1 degree Celsius
•
Mileage – Individual Tire (GPS Required)
• •
…the SST TPMS Sensor unit has proven to be the most accurate and robust sensor on the market
TPMS - ROI •
Customers value the environmental and economic benefits of TPMS but the safety benefits drive their demand…(Commercial)
•
Key value proposition: reduced fuel consumption, reduced annual tire cost, reduced downtime and road service expense…TPMS has now become a production tool…(OTR, Mining)
SST TPMS Product Categories OTR (Mining) Closed System – Harsh Environment Public Transit (Bus) Interface to Network – Low Profile Municipal and Commercial Waste Haulers GPS Integration - Driver Alert capabilities
TPMS – OTR and Commercial “Real World” •
Our Lab - Highland Valley Copper Kamloops, BC
•
Our own Ex Whatcom County 44 passenger Orion Mark 5 transit bus
A few of our Customers: - NACG (North American Construction Group) - Syncrude (Alberta, Tar Sands) - Copper Mountain Mines Inc, - Klempke (Alberta, Tar Sands) - King County Transit - Edmonton Transit - Edmonton Waste - Vancouver Waste
General IMU microsystem design strategy •
Present generation of automotive MEMS sensors – suitable for air bag systems, vehicle stability systems, etc., but not enough performance/low cost ratio for inertial navigation
•
Estimated resolution (input from Melexis and SST Wireless):
•
•
Linear accelerations (ax, ay): 0.001g in 40Hz bandwidth
•
Angular rate (Ωz) sensing: 0.01deg/sec in 30Hz bandwidth
Design strategy: operate the microdevices in a strong (nonlinear) electromechanical coupling regime -> advanced digital signal processing algorithms in FPGA, using analog CMOS as a thin interface.
System level view
Global challenges and ways of managing them •
Complex microsystem: combination of MEMS devices (accelerometers and gyroscopes), CMOS analog electronics and digital signal processing implemented in reprogrammable hardware
•
Different design tools and simulation levels: finite elements (for MEMS and packaging), Spice, VHDL/Verilog, Simulink/Matlab
•
How do they fit together? •
Extract reduced order behavioral models in VHDL-AMS/Verilog-A for the inertial MEMS devices, to be used in the design and simulation of the analog CMOS subsystem
•
Simplified behavioral models (from simulations or experimental measurements) for the subsystems, to use them in Simulink system level architecture
•
Make possible (through design) the independent test of the subsystems
Inertial MEMS sensors •
Challenges: higher resolution and accuracy in a small die area, taking into account environment (e.g. temperature) fluctuations
•
The intrinsic mechanical sensitivity ~ inertial mass => bad scaling!!
•
Exploit innovative nonlinear coupling effects: operation on the stability border for accelerometers, parametric amplification for gyroscopes
•
Closed loop configuration: thin analog layer (capacitive sensing and actuation), with most of the feedback control in the digital domain
•
Silicon-on-insulator fabrication technologies: MEMSCap SOI-MUMPs, Tronics (fabrication services subsidized by CMC Microsystems)
High-sensitivity accelerometers (1) •
Pull-in based MEMS accelerometers – operation on the stability border, in order to get micro-g resolution
•
Collaboration with Delft University of Technology, The Netherlands, and University of Porto, Portugal
•
Fabrication technology: SOI-MUMPs (MEMSCAP)
Operating principle
•
Exploit the high sensitivity of the pull-in time to external forces in electrostatically operated MEMS devices -> resolution set primarily by the resolution of the time measurement
Pull-in accelerometer performance
a=
Vstep V pi
•
Calibrated measurements (using a reference optical fiber Bragg accelerometer) have shown a resolution of 0.25us/ug, in a measurement noise floor of about 400ug (intrinsic mechano-thermal noise ~ 40ug for a bandwidth~2/tpi = 180Hz)
•
Simple electronics for a->time conversion, but nonlinear dependence => conversion table or digital interpolation necessary
•
Challenge: the damping coefficient (and implicitly tpi ) strongly depends on the operating temperature => self-calibration needed in operation!
Publications – pull-in accelerometer 1.
R.A. Dias, L. Mol, R.F. Wolffenbuttel, E. Cretu, L.A. Rocha [2011]- “Design of a time-based micro-g accelerometer,” in IEEE Sensors, in press
2.
R.A. Dias, E. Cretu, R.F. Wolffenbuttel, L.A. Rocha [2011]”Pull-in based ugresolution accelerometer: characterization and noise analysis,” Sensors and Actuators A: Physical, Feb 2011
3.
L.A. Rocha, R.A. Dias, E. Cretu, L. Mol, R.F. Wolffenbuttel [2011]”Auto-calibration of capacitive MEMS accelerometers based on pull-in voltage,” in Microsystem technologies, vol. 17, no. 3, pp. 429-436, Feb 2011
4.
R. Dias, E. Cretu, R. F. Wolffenbuttel, and L. Rocha [2010] “Characterization of a Pull-In Based µg resolution Accelerometer,” in Eurosensors XXIV
5.
R. A. Dias, R. F. Wolffenbuttel, E. Cretu, and L. A. Rocha, “Squeeze-film damper design with air channels: experimental verification,” in Eurosensors XXV, Athens, Greece, 2011, Sep 4-7
High-sensitivity accelerometers (2) – Closed-loop case •
Goal: digital control loop, easy to implement in FPGA
•
Classic control methodology: sigma-delta
•
Novel MEMS control architecture: sliding-mode control – two-level actuation
From simulation to experiment and back •
Co-simulation with analog electronics and Simulink/FPGA ->needs behavioral (reduced order models) Polytec MSA-500 micromotion analyzer
Digital binary feedback accelerometer
Self-calibration principle for inertial MEMS sensors Principle: use binary pseudorandom sequences - electrostatic actuation with small amplitude PN to recover the impulse response Self-test and self-calibration – requirements from SST Wireless LTI system
Reference model self-calibration architecture Tuning Block (Fel)
MEMS Device DUT
Read-out
FPGA
DAC PN Generator
MEMS DUT-R
DAC
ADC
-
+
Vout Calibration
Adaptive self-calibration
Self-calibration FPGA blocks PN Sequence generator
DUT-R
Experimental tests
A. Kansal, E. H. Sarraf, M. Sharma, and E. Cretu [2011] “Novel Adaptive FPGA-based selfcalibration and self-testing scheme with PN sequences for MEMS-based inertial sensors,” in IMS3TW 2011, Santa Barbara, California, USA, May 16-18, 2011
MEMS gyroscope
Nonlinear amplification techniques in MEMS gyroscopes •
Electro-mechanical parametric amplification for vibratory angular rate sensors: modulate electrically the stiffness of the sense mode
MEMS gyroscope publications 1.
M. Sharma, E. H. Sarraf, and E. Cretu [2011] “Parametric amplification/damping in MEMS gyroscopes,” in IEEE MEMS 2011, The 24th IEEE Int. Conf. on Micro Electro Mechanical Systems, Jan 23-27, 2011, Cancun, Mexico.
2.
I. Sabageh, V. Rajaraman, E. Cretu, and P. J. French [2010] “Design and Modelling of a Decoupled and Tunable (40–330 Hz) SOI-MEMS Yaw Rate Gyroscope,” in STW SAFE 2010 - Semiconductor Advances for future electronics and sensors, 2010.
3.
I. Sabageh, V. Rajaraman, E. Cretu, and P. J. French [2010] “Design and modeling of a three-mass, decoupled, tunable SOI-MEMS gyroscope with sense frame architecture,” in 21st Micromechanics and Microsystems Europe Workshop (MME2010), 2010.
4.
M. Sharma, E. H. Sarraf, and E. Cretu, “A novel dynamic pull in MEMS gyroscope,” in Eurosensors XXV, Athens, Greece, 2011, Sep 4-7
Capacitive sensing CMOS circuit
Low-noise parasitic-insensitive switched-capacitor-based CMOS front-end Output voltage is proportional to input capacitance variation The output voltage is digitized by an on-chip analog-to-digital converter
Analog CMOS interface – learned lessons •
The interface between MEMS and CMOS dies is critical – do not underestimate the parasitic effects in the real world
•
Design for test and calibration! – the CMOS chip should be testable in the absence of any MEMS die.
•
J. Shiah, H. Rashtian, and S. Mirabbasi, " A Low-Noise Parasitic-Insensitive CMOS Switched-Capacitor Interface Circuit for MEMS Capacitive Sensors," to be presented at IEEE International NEWCAS Conference, June 2011.
•
Alternative solution for incipient tests with FPGA: lower-performance PCB capacitive readout
Kalman filtering for path prediction •
We use the square-root Kalman filter -> needs less numerical precision for the same performance
•
Kalman filter is computationally demanding requiring: •
QR Decomposition
•
Matrix inversion
System level: Automotive Reconfigurable Guidance System
Analog Part Accelerometer
Kalman Filter QR Decomposition with CORDIC
Instrumentation Amplifier
Matrix Inversion
A/D Converter
Path estimation
Accelerometer (UBC) + Instrumentation Amplifier (UVic + UBC) Kalman Filter (QR Decomposition, Matrix Inversion) on FPGA (UVic) Path estimation (UVic + UBC) + A/D Converter (UBC)
QR Decomposition- Accuracy in FPGA •
Matrix inversion (required by the Kalman tracking block) is expensive for realtime computation => made using QR algorithm, implemented through CORDIC schemes in FPGA
•
Number of test vectors = 105
•
Accuracy threshold = 10-3
•
Number of CORDIC iterations = 18
Custom- designed reconfigurable hardware
Commercial (fine-grained) FPGAs:
Large penalties (silicon area, power consumption)
Good for prototyping and low-batch implementation
Novelty: Coarse-grained FPGA: ShEERA (Shift-enabled Embedded Reconfigurable Array)
Based on shift-enabled interconnection network
Good for vector rotations, QR decomposition, etc.
Generic architecture for an Automotive Reconfigurable Computing Platform (AReComP)
Shift-Enabled Embedded Reconfigurable Array
Assembly and packaging – U. Alberta Flip chip Facility Specification • Three technique for bonding • Applied heat on the placing tool for the device and highly controlled bottom heater for the substrate • Friction welding using Ultrasonic module • Force and epoxy bonding for cold bonding • Placement Accuracy: within 5 µm in tolerance
Ultrasonic module thermo module
• Max. Board Size:410 mm x 234 mm with potential for upgrades • Chip Sizes: 0.2 mm up to 85 mm with upgrade options depending on the bonding technique • Bonding Force: min. 0.1 N max. 500 N • Total Magnification: up to 230 x
Dispensing module
Force module
Integration of MEMS Sensors in Flexible Materials Integration of MEMS Sensors in Flexible Materials Integration techniques investigated • Solder ball pickup and placement tool design: This is done using a homemade tool designed to pick different size of balls and micro assemble them on the substrate using the same level of tolerance the system can offer for the devices. • Gold ball bumps placement using wire bonding. Characterization of the bonding strength with variations in the bonding parameters. • Flip chip assembly on flexible substrates using ultrasonic frication. Side effect issues such as damping of the ultrasonic vibration is investigated using bump stacking. • Conductive epoxy and anisotropic conductive epoxy will be tested for flexible PCBs • Potential application • Gharib, H.H., Mohamed M. EL Gowini, Moussa, W.A., "Developing a MEMSBased Smart Tire", Auto21, Windsor (2010). [2nd place/78 participating poster] F304-FIS
Conclusions •
Large, complex project – it is essential to ensure by design a good interface between the subsystems and their independent testability
•
The same module will present different views/models, depending on the design level: layout, finite element model, behavioral model in Spice/VHDL-AMS or Simulink, etc.
•
Several innovative concepts: nonlinear amplification techniques for the MEMS devices, parallel CORDIC architecture for fast matrix inversion in FPGA and dedicated, low-cost, coarse-grained reconfigurable die (SHEERA), novel switched-capacitor interface, novel packaging techniques
•
Challenging but rewarding work – 8 HQPs, large publication number
•
The on-going collaboration with SST Wireless is essential, in order to get a realistic feedback “from the trenches” and for the wireless interface with other automotive modules.
