This presentation, "Synchronized Energy Harvesting Sensor Networks", was made at the 2009 Sensors Expo in Rosemount, IL. MicroStrain has developed the next generation energy harvesting wireless nodes for health monitoring of structures and condition based monitoring of machinery. These nodes can be integrated into a large scale network, controlled with a Wireless Sensor Data Aggregator (WSDA) providing node-to-node synchronization up to +/- 4 microseconds.

1. Synchronized Energy Harvesting
Sensor Networks
S.W. Arms*, J.H. Galbreath, C.P. Townsend,
D.L. Ch hill N
D L Churchill, Nam Ph ^
Phan^
* President, CEO ^Division Head (Acting) Structures
MicroStrain,
MicroStrain Inc Structures Division - AIR 4 3 3 2
4.3.3.2
Williston, Vermont Naval Air Systems Command
www.microstrain.com Patuxent River, Maryland
swarms@microstrain.com.
Proc. Sensors Expo, Rosemount, IL, June 9h, 2009 © microstrain, inc. 2009 all rights reserved

2. Sensing
the Future
Wireless sensors, in the billions, will become
deeply embedded within structures &
machines.
Sensed information will be automatically
compressed & forwarded for condition
based maintenance.
© microstrain, inc. 2009

3. The
Economist
April 28th – May 4th
2007
“We’re wearing out…plan to replace us soon”

4. Problem:
But h
B who will replace all
ill l ll
those dead batteries?
© microstrain, inc. 2009

5. Solution:
• Harvest & store energy from strain
strain,
vibration,
vibration, motion, thermal gradients,
light, electromagnetic fields
light
• Use power management to balance the
energy “checkbook”
“checkbook”
• Use embedded processors to compress
data,
data compute fatigue life
© microstrain, inc. 2009

6. Bell M412 Installation
Strain energy
harvesting wireless
pitch link installed on
Bell M412 Feb 2007
(1st time ever)
Patents pending © microstrain, inc. 2009

7. Flight
Test
Results
• Passed
– in-flight EMI evaluations
in-
– rotor track & balance verification
• Data were collected wirelessly on board the
aircraft with no indication of data loss
Patents pending © microstrain, inc. 2009

8. Objective: Demonstrate a synchronized,
energy harvesting, wireless structural health
harvesting
monitoring & reporting system for helicopters
© microstrain, inc. 2009
Patents pending

9. Detailed
D t il d
Objectives
• Develop a wireless data aggregator (WSDA), capable of
synchronizing wireless/hard-wired sensor networks and
wireless/hard-
aggregating data with open architecture communications
to HUMS
HUMS.
• Document time synchronization accuracy
• Develop a high sample rate wireless sensor node for
helicopter gearbox apps.
• Demonstrate system compatibility with a scalable
network of active RFID
t k f ti RFIDs.

10. Communicatingg
Wirelessly
y

11. MicroStrain’s Wireless Sensor Networks
(IEEE 802.15.4)
( 802.15.
Time Division Multiple Access Frequency Division
(
(TDMA) &
) Multiple Access (FDMA)
Carrier Sense
Multiple Access (CSMA)
© microstrain, inc. 2008

12. How many nodes will this
low power TDMA system
support?
Aggregate sample rate (Hz):
~10,000/total no. sensor channels
10 000/total no
i.e.: ~100 single ch nodes can
i 100 i l h d
transmit data at 100 samples/sec
Patents pending © microstrain, inc. 2008

13. Methods:
Timing Engine

14. Previous Work
• Le Cam, V., “Synchronization of Wireless Sensors:
Review of Methodologies, Experience Feedback of the
Very Precise GPS Solution”, Third European Workshop
Solution
on Structural Health Monitoring, July 5-7, Granada,
5-
Spain, July 5-7, 2006
5-
Placed GPS receivers at each wireless node
to achieve absolute precision of 1
microsecond

15. Data Aggregator collects
gg g
time synchronized data
w/ 4 GB removable flash memory
All wireless nodes use precision nano-power real time clock
p p
(RTC) with +/- 3 ppm (-40 to +85 deg C) time reference.
Wired inertial sensor uses same time reference as Data
Aggregator.
Data Aggregator’s RTC uses Global Positioning System
’ C Gl b l S
(GPS) as time reference. Data Aggregator sends beacons to
update time sensing node’s time keepers
p g p
Patents pending © microstrain, inc. 2009

16. Timing Engine Overview
• Timing
Ti i engine provides th
i id the
following:
GPS
• 1pulse per second Receiver
CAN
CAN Nodes
CAN Nodes
CAN Nodes
CAN Nodes
(PPS) Synch Clock Controller
• Trigger Line Timing
Ti i
Wireless Wireless Nodes
• Extremely accurate Engine
Controller
Wireless Nodes
Wireless Nodes
Wireless Nodes
absolute time keeper
• The 1 PPS synchronization
y Wireless Wireless Nodes
Wireless N d
Wi l
clock is distributed to CAN Controller Wireless Nodes
WirelessNodes
Nodes
and 802.15.4 network
802.15.
controllers. (green) µP Core
running
Linux USB Node
SYNCH CLK & TRIG
• In turn, network controllers
turn USB N d
Node
CAN Synch Mechanism
propagate the 1PPS clock to
nodes through a high-priority
high- RS-232 Node
Wireless Synch Beacon
broadcast beacon packet. RS-232 Node
(blue and orange)
Patents pending © microstrain, inc. 2009

17. Harvesting
g
Energy
gy

18. Vibration vs. Strain Energy Harvesters:
gy
Gearbox Resonant Energy
G b R tE
Harvester Output: 37 mW
Volume: 4.3 cc
Weight: 38 grams
Flexible Strain Harvester Output:
~14 uW per sq cm (90 uW per sq in)
@ 200 uE p-p, 4.3 Hz
p p,
On Bell 412 Pitch Link 12 patches
delivered ~ 1 uW/lb (200-400 uW)
Weight: 4.3 gr/patch*12 patches = 52 gr
Patents pending © microstrain, inc. 2009

19. Sensing strain
strain,
force, pressure,
force pressure
torque, vibration,
torque vibration
temperature
Wirelessly

20. MicroStrain’s embedded firmware
optimized for strain gauges
• Wireless offset adjust
• Wireless gain adjust
• Wireless co t o o sa p e rates
e ess control of sample ates
• Wireless shunt cal – bits to microstrain
• Low tempco’s:
tempco s
tempco’s:s:
offset: -.007%/C , span: .015%/C
015%/C
• Mux’d, pulsed & regulated bridge excitation
Mux dd,
Patents pending © microstrain, inc. 2009

21. Wireless Pitch Link Strain
& Load Sensing Nodes
Fractal antennas
Shear-Link ™
© microstrain, inc. 2009
Patents Pending

22. Pitch Link Consumption for Various
Operating Modes:
O ti M d
• Mode 1: Wait until stored energy crosses threshold: nanoamp
threshold:
comparator turns circuit “on”. Predetermined amount of data
transmitted. Consumption varies with available energy, timekeeper
draws 9 microwatts.
• Mode 2: Data logged to memory: Download at end of test.
Consumption @ 32 samples/sec: ~100 uwatts
• Mode 3: Transmit if energy allows: Log 100 samples, check stored
energy, transmit if possible. Consumption with 32 samples/sec: ~250
uW,
uW, drops to 100 uW without radio transmission.
• Mode 4: Real Time Transmission: Log 100 samples, then transmit.
Consumption with 32 samples/sec: ~250 uwatts.
uwatts.
© microstrain, inc. 2009
Patents pending

23. Wireless Bridge (Strain)
System Di
S t Diagram
Time
keeper
(patents pending)
© microstrain, inc. 2009

24. High Speed Wireless Node
• Programmable sample rates,
g p
offsets,
offsets, gains, & anti-aliasing
anti-
filters
• A/D resolution: 16 bits
• 100 KHz A/D sampling rate (3
ch,
ch, simultaneous, full diff)
• Event consists of 125,000
samples (or 0.4 seconds at
100 KHz sample rate)
• Stores 1 million samples on 2
MB embedded, non-volatile
memory
Patents Pending © microstrain, inc. 2009

25. Energy
Conservation:
work to balance the
“energy checkbook”
“ h kb k”
© microstrain, inc. 2009

26. Powering down between
g
samples greatly reduces
power consumption
Patents pending © microstrain, inc. 2009

27. Embedded routines allow
microelectronics to
adapt to
the
th amount of available
t f il bl
energy
© microstrain, inc. 2009
Patents pending

28. Average power consumption
g p p
(mW) for 50 kSPS data
mW)
acquisition
i iti
Acquisition Interval
Sample 1 min 10 min 1 hour 1 day
y
duration
(sec)
0.1 1.67 0.22 0.09 0.06
0.5 8.09 0.86 0.19 0.07
1.0
10 16.1
16 1 1.66
1 66 0.33
0 33 0.07
0 07
© microstrain, inc. 2009

29. Results:
Wireless Network Timing -
How quickly are broadcast
commands processed?

30. Broadcast Synchronization
Results (4 nodes)
• Waveform at right shows
g Scope traces
4 captured waveforms
representing the start of
sampling for each of the 4
p g
nodes.
• Note that each node
starts sampling at slightly
different times. In this
specific case the last
case,
node started sampling
~3.4 microseconds (µsec)
after the first node
node.
© microstrain, inc. 2009

31. Broadcast Synchronization
Results
Scope traces
• After repeating the
broadcast trigger
command 250 times, the
d ti th
timing differences are
bound within an envelope
of ±4 µsec. This
represents the initial
synchronization accuracy
for the group of nodes.
© microstrain, inc. 2009

32. Results:
Wired & Wireless Network
Timing:
How well synced w/
beaconing?
g

33. Saw tooth Input, room temp.
2 nodes, 2 h
d hours @ room t
temperature,
t
clock drift: ~325 us (45 ppb) between the two sensor nodes
Timing beacon sent once – at start of test only.
Node 1 & Node 2
Sensor Data Overlay
3
2.5
ensor Input (V)
2
1.5
1
Se
0.5
0
7199.94
7199 94 7199.96
7199 96 7199.98
7199 98 7200.00
7200 00 7200.02
7200 02 7200.04
7200 04 7200.06
7200 06 7200.08
7200 08
Time (seconds)
© microstrain, inc. 2009

34. Saw Tooth Input, -40 to +85 deg C
2 nodes 2 hours w/ 10 Hz Sawtooth: clock drift 5.71 msec (793 ppb)
nodes, hours, 5 71
w/ 1 Hz Sawtooth: clock drift 5.04 msec (700 ppb)
Timing beacon sent once – at start of test only.
Node 1 & Node 2
Sensor Data Overlay
3
2.5
Sensor Input (V)
2
1.5
1
0.5
0
7199.70 7199.72 7199.74 7199.76 7199.78 7199.80 7199.82 7199.84
Time (seconds)
© microstrain, inc. 2009

35. Timing Results Summary
• Synch beacon sent once - at start of test only - provided
~5 ms timing accuracy over 2 hours, subjected to -40 to
+85 C.
• Synch w/ periodic (60 sec) beacon provided +/- 50 us
timing accuracy over 13 hours, subjected to -40 to +85 C.
g y , j
• Conservative approach: send resync beacon every 5
minutes to achieve sub millisecond timing accuracy when
sub-millisecond
temperatures are extreme and changing rapidly.
© microstrain, inc. 2009

36. Conclusions
• Accurate time synch developed for wireless
sensor nets that doesn’t require GPS.
• System supports high sample rate (50 KHz)
sensor nodes, and active RFID tags
• Provides open architecture interface to HUMS
to eliminate wires and enable reductions in
weight and complexity.
Patents pending © microstrain, inc. 2009

37. Conclusions (continued)
• High sample rate nodes can operate
perpetually, without batteries, from gearbox
vibration alone.
• Supports remote reporting over mobile phone
networks (satellite reporting currently under
development).
development)
• These capabilities, coupled with appropriate
wireless security methods, will enable critical
structural sensor data to be managed remotely,
securely, and automatically.
Patents pending © microstrain, inc. 2009

38. Need More Info?
• Sensors Expo Booth 1005
• www.microstrain.com
• www.microstrain.com/customer-docs.htm
www.microstrain.com/customer-
• swarms@microstrain.com
@

39. Acknowledgements:
Navy/NAVAIR SBIR PH II
ONR BAA
Bell Helicopter
p
Thank You!

40. References:
• M.J. Hamel et al., Energy Harvesting for Wireless Sensor Operation and Data
Transmission, US Patent Appl. Publ. US 2004/0078662A1, filed March 2003
• D L Churchill et al., Strain Energy Harvesting for Wireless Sensor Networks,
D.L. Ch hill l S i E H i f Wi l S N k
Smart Structures and Materials, SPIE, vol. 5005, pp. 319–327, 2003
• S.W. Arms et al., Shaft Mounted Energy Harvesting System for Wireless
g g
Sensor Operation and Data Transmission, US Patent Appl. Publ. US
d l bl
2005/0017602A1, filed Jan 2004
• S.W. Arms et al., Wireless Strain Measurement Systems for Aircraft Test,
, y Test,
Aerospace Test Expo, Anaheim, CA, Nov 2006
• S.W. Arms et al., Energy Harvesting Wireless Sensors for Helicopter Damage
Tracking, American Helicopter Society Annual Forum, Phoenix, AZ, May 2006
g, p y , , , y
• S.W. Arms, C.P. Townsend, D.L. Churchill, M. Augustin, D.Yeary, P. Darden,
Augustin, D.Yeary,
N. Phan, Tracking Pitch Link Dynamic Loads with Energy Harvesting Wireless
Sensors, AHS 63 d Annual Forum, Virginia Beach, VA, May 2007
S 63nd A lF Vi i i B h VA M