Ultra-Low-Power Integrated Circuits and Physiochemical Sensors for Next-Generation “Unawearables” Patrick Mercier University of California, San Diego
Source: Cisco 2
Wearables: an exciting high-growth market Medical 3 billion wearables Fitness shipped by 2025* Infotainment Industrial Source: Transparency Market Research *IDTechEx 2015 Report
Why aren’t we there now? Size & Usability: Need to develop sensors that are small & seamlessly integrated into daily life Mission: Battery Life: Address these issues through Need ultra-low-power and/or energy harvesting innovative to minimize re-charging transdisciplinary research Utility: Need to develop sensors that are actually useful 4
Wearables Roadmap PHASE I PHASE II PHASE III Devices on the wrist Patches Unawearables + Well-understood use case + Many sensing modalities Applications: + Room for a battery MARKET SIZE Health & Fitness – Requires daily user interaction – Limited sensing Entertainment – Not convenient opportunities Medical – Maintenance burden + Built directly into already used objects/textiles + Many sensing modalities + Automatic wireless comms, energy harvesting Research challenges: new biosensors, ultra-low-power bioelectronics, energy harvesting, soft integration 2020 2022 2018 5
Why aren’t we there now? Size & Usability: Need to develop sensors that are small & seamlessly integrated into daily life Mission: Battery Life: Address these issues through Need ultra-low-power and/or energy harvesting innovative to minimize re-charging transdisciplinary research Utility: Need to develop sensors that are actually useful 6
Wearable sensing opportunities Physical attributes Electrical attributes • Motion (e.g., steps) • ECG (heart) • Temperature • EEG (brain) • Respiration • EMG (muscles) 7
Electrophysiology today Wet electrodes: Non-contact electrodes: • Inconvenient • Very convenient (can integrate into textiles) • Irritating • Opportunities for large number of channels • Good performance • Severe motion artifacts Cognionics G. Cauwenberghs 8
Hardware + software co-design for motion artifact reduction Experimental results: Naïve solution: Measure electrode motion Sitting via accelerometer Problem: measures absolute motion; not motion w.r.t. body Proposed solution: Dynamically measure change in electrode impedance via a dual-channel electrode Walking Cc(t) Vs(t) V1(t) x1 To electrode1 Cino Cp1 Cc(t) Vs(t) V2(t) x1 To electrode2 Cino Cp2 Up to 76% reduction of artifacts Lin et al., BioCAS 2015 9
Fully-on-chip Wireless Neural Interfacing Devices m m ADVANTAGES: 4 Pt Electrode V DD_STM Voltage [V] Adiabatic current Measured • Fully-integrated: no wires, batteries, or any other 2 V STM_U stimulator: external components 0 V STM_L 6x more efficient than • • Fully encapsulated with biocompatible material: no 35.7 k W 1.1 k W V SS_STM -2 W conventional approaches adverse reactions with the brain 150 >2x more efficient than • • Microchip integration means upwards of 100s of 100 Current [ m A] Stimulation 33 nF 50 I STM_U m prior work that use large channels per chip 0 I STM_L -50 off-chip inductors • Completely modular design -100 -150 • Possible to place many chips in the brain for large- 0 1 2 3 4 5 6 7 scale recording/stimulation Time [ms] S. Ha et al., VLSI’15 / TBioCAS’18 (d)
Strain sensing for detecting risk of fibrosis in head+neck cancer patients J. Ramirez et al., ACS Nano, 2018 11
Machine learning for classification 84% accurate model J. Ramirez et al., ACS Nano, 2018 12
Wearable sensing opportunities Most of the wearables market today Physical attributes Electrical attributes • Motion (e.g., steps) • ECG (heart) • Temperature • EEG (brain) • Respiration • EMG (muscles) • Blood pressure Biochemical attributes • Glucose Opportunity! • Electrolytes • Alcohol • Lactate • Many more! 13
Biochemical Sensing Today Conventional lab testing • Expensive, painful, time consuming/inconvenient • Very infrequent spot measurements Point-of-care devices • Often still needs access to blood (invasive) • Infrequent spot measurements (subsampling) Research need: non-invasive, continuous measurement devices
Example: lactate monitoring for athletes Staying below the “lactate Current state-of-the-art threshold” important for testing method: endurance training Non-invasive and/or continuous sensing is required 15
Hybrid physiochemical & electrophysiological sensing Current ( µ A) 3 Lactate 2 1 0 200 H. R. (B/min) 160 120 80 First demonstration of 0 500 1000 1500 2000 Time (s) simultaneous chemical+electrophysiological Opportunities for sensing in a wearable patch data analytics! S. Imani et al., Nature Communications, 2016
Hybrid physiochemical/electrophysiological sensor operation S. Imani et al., Nature Communications, 2016
Non-invasive wearable alcohol sensor Epidermal prototype: Electrochemical analysis after iontophoresis: To induce sweating à capture ethanol at the skin surface Measurement procedure: YOU ARE DRUNK! J. Kim et al., ACS Sensors, 2016 18
Non-invasive dual-fluid glucose/alcohol sensing A wireless “glucohol” sensing platform J. Kim et al., Advanced Science, 2018 19
A wireless saliva sensor in a mouthguard Health applications Fitness applications Measure Uric Acid for Measure Lactate for Hyperuricemia Stress / Exertion Startup company: J. Kim et al., Biosensors & Bioelectronics, 2015 20
Why aren’t we there now? Size & Usability: Need to develop sensors that are small & seamlessly integrated into daily life Mission: Battery Life: Address these issues through Need ultra-low-power and/or energy harvesting innovative to minimize re-charging transdisciplinary research Utility: Need to develop sensors that are actually useful 21
Major limiter: battery size / battery life Power breakdown: Processor, Power 1% management, 1% Sensors, 18% Radio, 80% Battery Research goal: Minimize power of load circuits (especially RF), and perform energy harvesting G. Burra et al., ULP Short-Range Radios (Mercier & Chandrakasan, Eds.), Springer’15 22
Near-zero-power RF transmitter Direct-RF 2.4GHz Power 4Mbps OOK Active power: Oscillator w/ 2.8mm loop antenna 154μW Sleep power: 500pW Average power: 2.4nW If TX power can be so low, the power consumption of even basic building blocks begins to matter Start-up time < 52ns H. Wang et al., JSSC’18 23
Ultra-Low-Power Voltage Reference Generator Normalized Reference Voltage Reference Voltage V REF [mV] 1.01 Sub-pW power 348 consumption! 1.005 346 1 344 0.995 0.99 342 Inspired by Seok et al., JSSC’12 -20 0 20 40 60 Temperature [ o C] H. Wang et al., Sci. Rep.’17
A 420fW self-regulated 3T voltage reference generator 9x improvement Power: 420fW Temperature coefficient: 252ppm/°C (N=38 chips) Line regulation: 0.47%/V H. Wang et al., ESSCIRC’17
A 3.4pW 5T current reference generator Conventional Proposed • Eliminates additional amplifier bias current • Inherent self- regulation • Gate-leakage transistor to reduce area H. Wang et al., SSCL’18 26
pW relaxation oscillator pW current reference H. Wang et al., Sci. Rep.’17 27
pW relaxation oscillator: 65nm test chip results 214 1.03 45 40 212 1.02 35 Normalized Frequency 210 1.01 Frequency [mHz] 30 Power [pW] 208 1 25 206 0.99 20 204 15 0.98 202 10 0.97 5 200 -20 0 20 40 60 -20 0 20 40 60 Temperature [ o C] Temperature [ o C] H. Wang et al., Sci. Rep.’17 28
pW temperature sensor H. Wang et al., Sci. Rep.’17 29
Temperature sensor measurement results 65nm prototype 1.5 1 Worst-case inaccuracy [ o C] Temperature Error [ o C] 0.5 645x lower power 0 -0.5 -1 -1.5 -20 0 20 40 60 Temperature [ o C] Power [nW] Consumes only 110pW with +/- 1.9 ℃ inaccuracy New design with better performance under review H. Wang et al., Sci. Rep.’17 30
Sub-nW SAR ADC BIT[9:0] Control C DAC Signal C SH 1 MHz CLK 10 Hz CLK Power: 780pW Sampling rate: 10 S/s ENOB: 8.3b H. Wang et al., JSSC’18 31
Power Management Unit o 1.8V battery to 0.6V load conversion via a 3:1 Dickson topology o Minimized leakage power and high SSL metric o Non-overlapping clock reduces quiescent power by 21% o Peak efficiency: 96.8% at 100nA, 10Hz H. Wang et al., JSSC’18 32
A 5.5nW Wireless Ion-Sensing System Average power consumption: 5.5nW H. Wang et al., JSSC’18 33
Recommend
More recommend
