Sensors and batteries are the most important constituents of a wearable device. They can make or break the entire IoT system. Typically, a designer of wearable electronics pays less attention to the important issues of sensors’ performance, batteries, power management etc, during the design process. These issues are indeed fundamental to the success of a wearable device. The half-day workshop covered fundamental design concepts revolving around sensors, calibration, batteries and power management. The instructors shared insights which they have gained through extensive industrial research and technology management.

1. IEEE IoT/Wearable Technology Tutorial, 2017
7th Sep 2017, 10 AM to 1:00 PM, Hotel Hilton, Bangalore
www.inertialelements.comwww.oblu.io
Amit K Gupta
Founder & CEO, oblu.io
Wearable Electronics
A Designer’s Perspective
Subhojyoti Bose
Research Engr, oblu.io
Amey Karkare
Assoc Prof, IIT Kanpur

2. Outline
• Power Management
• Batteries
• Sensors
• Calibration
• Laboratory exercise
– Inertial Sensors’ calibration

4. Power Management
• Voltage supply / Power source
• Non-idealities, imperfections
• Power rating
• Real life design challenges
• Efficiency
• USB power supply
• Voltage regulator

5. Vout = ?
Question 1

6. Vout = ?
Question 1

7. Vout’ = ?
Vout’, IL (RL→0) = ?
Vout’, IL (RL→∞) = ?RL
IL
Question 2

8. • What is an ideal DC voltage source / power
supply ?
• What are the common non idealities of a typical
DC voltage source ?
• Now go back to Q2 and give reason for drop in
Vout with load.
Question 3

9. Vout’ = ?
Vout’, IL (RL→0) = ?
Vout’, IL (RL→∞) = ?RL
IL
Question 2

10. • Why there is drop in voltage during peak
summer / winter ?
• Why are you supposed to increase your meter
wattage if you install an AC in your house?
Question 4

11. Vout’’ = ?
Vout’’, IL (IL→0) = ?
Vout’’, IL (IL→∞) = ?
IL
IL
Question 5

12. • You receive fixed AC voltage 240V (Ideal scenario)
• What about “current”. Is current also fixed?
• Who decides the upper limit of current?
• What parameters help in deciding the upper limit
of current?
AC Supply @ Your Home

13. • AC to DC conversion
• Cellphone charger is an excellent example
• SMPS of your PC: Accepts 240V AC input and gives
out +5, -5, +12, -12, 3.3V
– Is this information about an SMPS enough?
Power Handling in Electronic Circuits

14. Current rating is very important !
Power Rating of an SMPS

15. Vin
Iin
Charger / TV / PC
Input Power = Vin x Iin
Power Consumption

16. Vout
Icharging
Vin
Iin
Efficiency (Ƞ) =
Power In
Power Out
Cellphone
Charger
Cellphone Charger

17. VUSB = ?
AC ~240V DC / AC ?
VUSB = ?
Wall Charger Car Charger
Q: Is there any limit on the current carrying capability of a USB i/f?
Cellphone Charger

18. • From where does the car charger derive its
power?
• DC / AC ?
• Battery chemistry?
• Terminal voltage?
Car Charger

19. Q: What is the power efficiency of the charger if the
charging current here is 600 mA ?
VUSB
VIN
IIN
Efficiency of a Cellphone Charger

20. ABC GT Silicon
Q: What is the power efficiency of GT Silicon’s charger in
this case ?
VIN
IIN
VUSB
IUSB
Benchmarking

21. • Provides voltage at certain level (1.8V, 2.5V,
3.3V, 5V etc)
• Maintains the voltage despite variation in input
voltage, output current, temperature etc
• The max load current, maximum input voltage
range and ambient temperature range for
stable output are important specifications
Voltage Regulator

22. (Vin,min, Vin,max) Vout
Iout,max
Ƞ
Specifications of a Voltage Regulator

23. Specifications of CAT6219 (2.85V, 500mA)
Dropout characteristic Line regulation
Load regulation Vout vs Temp

24. • Voltage supply / Power source
• Variation in voltage due to current withdrawn
• Non-idealities, imperfections of a voltage supply
• Current rating (typical, max) of a system
• Real life design challenges due to imperfections
• Efficiency of a power mgmt unit
• USB power supply (current limit)
• Voltage regulator
– Stability of output voltage, Range of input voltage
– Efficiency, Effect of ambient and operating conditions
Summary

26. Batteries
• Battery chemistry
• Non idealities
• Li-polymer battery
• Charging & Discharging
• Series & Parallel combination
• Protection circuit

27. • Application decides the primary source of power
• Non portable appliances use Mains as the main power
source and battery for backup
– Your TV / PC are not portable items. Those are meant to
operate on main supply. Battery/inverter is used as backup
power. (Give more examples)
• Portable appliances use battery as the main power source
and mains for charging or backup
– Oblu (wearable sensor) runs on battery. It uses USB power
for charging battery and also as an alternate power source.
(Give more examples)
Power Source: Mains / Battery ?

28. • AC Mains (Converted to DC internally)
• Batteries
• USB (?)
Typical Power Source for Electronic Gadgets

29. • Fixed voltage – 5V DC
• Maximum current limit of a USB port (which also supports USB
data transfer)
– 100 mA
– 500 mA
• Cellphone chargers use USB connectors
• Those are just physical connectors for charging only
• There is no USB data transfer support in chargers
• Charger circuit controls the charging current
• Charging current is chosen as per battery specifications
• Use only the charger specified by phone manufacturer !
Image source: Internet
USB Port as Power Source (?)

30. Image source: Internet
How would you explain this to a 2 yrs old

31. Image source: Internet
Phone is eating food!

32. Image source: Internet
Do these appliances also eat food?

33. They do! But they don’t have stomach!
Image source: Internet
Do these appliances also eat food?

34. • A perfect analogy to understand
• Form of energy – Chemical in both the cases
• Source of energy
• Energy storage
• Charge rate and time
• Discharge rate and time
• Energy draining beyond a limit
• Recovering some energy after some rest
• Recovering from a fatal state
Batteries and Human Beings

35. • Commonly used rechargeable batteries
– Li-ion / Li-Poly (Lithium ion / Lithium polymer)
– Pb-Acid (Lead Acid)
– NiCd (Nickle Cadmium)
– NiMH (Nickle Metal Hydride)
• Different chemistries, different terminal voltages
• Somewhat similar characteristics (like humans)
• Li-ion / Li-Poly: most popular for portable and wearable IoT
– Highest energy density
– Low maintenance
– Ease of handling
Reference - http://batteryuniversity.com/learn/archive/whats_the_best_battery
Types of Batteries

36. • Small size
• High energy density
• Low price
• Longevity is least bothered
Important Factors for Wearable / Portable

37. • Typical terminal voltage of a unit cell (3.7V)
• Battery capacity (milli-Amp-Hours or mAH or C)
• Charging current
– Recommended C/2 for best performance
– Charging time with C/2 is ~2 hours
• Fast charging (2C, max limit)
– Typical is 2C
– Max limit of charge current
– Must not be used on regular basis
Li-ion Battery

38. Protection Circuit Module
• Over charging protection voltage (~4.2 V)
• Over discharging protection voltage (~2.7 V)
• Max discharging current protection
• Over current protection
• Short circuit protection

39. Charging Profile
• Terminal voltage profile
– Varies nonlinearly. Faster variation near empty and slower at
near completion
– Approximated as linear for indication purpose
– Typically varies from 3.2V (full discharge) to 4.2V (full charge)
Constant
Current
Constant
Voltage
Image source: Internet

40. Discharging Profile
• Cutoff voltage (3.0 V) at room temperature
• Faster discharge results in reduced capacity
• Charging cycles of a battery are limited
C
Image source: Internet

41. • OCV or EMF (Li-ion): 2.7
(min), 3.7 (typ), 4.2 (max)
• Rint: ~500 mOhms
• Vbat = EMF – Iload * Rint
• Non idealities can be modeled as internal resistance
• Remember momentary current surge of 1A can
cause terminal voltage drop by 0.5V !!
• Internal resistance limits battery backup of a
system!
Internal Resistance

42. Series Combination
• For increasing the terminal voltage (V, 2V, …)
• Results in increased internal resistance (r1+r2+…)
• Capacity remains unchanged of the resultant
battery (C)
• Use batteries from the same manufacturer, same
model
EMF = ∑ EMFi Rint = ∑ Rint,i

43. Parallel Combination
Capacity = ∑ Capacityi 1/Rint = ∑ 1/Rint,i
• For increasing the capacity (C1+C2+…)
• Results in reduced internal resistance (r1||r2||…)
• Terminal voltage remains unchanged (V)
• Use batteries from the same manufacturer
• Defect in any one battery of the combination gets
distributed in the resultant battery system

44. • Tendency of battery to gain some strength given
rest from the normal operation.
• Measure terminal voltage when in operation. Stop
using it for few minutes. Measure the voltage
again.
• Fully drained out cellphone also becomes alive for
few minutes, after sometime
Elasticity (The ability to self recover)

45. • Form factor, energy density, charge cycles
Cylindrical Prismatic Coin Cell
Image source: Internet
Battery’s shape

46. Image source - http://spectrum.ieee.org/tech-talk/consumer-electronics/portable-devices/ces-
2017-panasonic-shows-off-bendable-lithiumion-battery-for-iot-wearables
Flexible Li-ion Batteries for IoT, Wearables

47. Image source: www.amazon.in
Universal Battery Charger

48. We were shocked when we measured thickness of this coin cell
using this vernier caliper, without much thinking. Can you tell why
did it happen?
Flex Your Neurons

49. Flex Your Neurons
What is this?

50. Electrodes (Cathode, Anode)
Insulator
Li-ion battery unpacked!
Flex Your Neurons
What is this?

51. • Similar to human beings
• Energy storage (food)
• Internal resistance (weakness, immunity)
• Li-poly rechargeable battery
– Capacity 3.7V XXXXmAH (milli Ampere Hours)
• Terminal voltage’s variation with charge/discharge
• Max. charging & discharging rates (2 hrs charging)
• Serial & parallel combinations
• Protection circuit – Min and max cut-off voltages
Summary

53. Sensors
• Technology Trend
• Sensors’ characterization
• MEMS sensors:
• Pressure Sensor
• Accelerometer
• Gyroscope
• A case study of a motion sensing application
• Shoe-mounted inertial navigation

54. • A sensor typically measures or identifies a particular
physical quantity.
• Sensors convert the physical properties to electrical
signals understandle by machines.
• Sensors are found everywhere.
Image source: Internet
Sensors

55. Changing Trends in 21st Century
Image source: Internet

56. Wearable Devices
Image source: Internet

57. Internet of Things

58. • Touchscreen
• Light
• WiFi
• Wind speed
• Bluetooth
• GPS
• Proximity
• Barometer
• Tilt
• Magnetometer
• Accelerometer
• Gyroscope
• Temperature
• Humidity
What is a sensor ?
Smartphone – A Sensor Hub
Image source: Internet

59. Is watch a sensor?
Time Sensor??
Image source: Internet

60. Trend Enabling Technology
VLSI Technology & Moore’s Law
The observation made in
1965 by Gordon Moore,
co-founder of Intel, that
the number of transistors
per square inch on
integrated circuits had
doubled every year since
the integrated circuit was
invented. Moore predicted
that this trend would
continue for the
foreseeable future.
Source: Wikipedia Image source: Internet

61. Trend Enabling Technology
Micro Electro Mechanical Systems (MEMS)
Accelerometer Gyroscope
Image source: Internet

62. Micro Electro Mechanical Systems
• Miniaturized mechanical and electro-mechanical elements
• Moving structures fabricated on a Silicon substrate
• Made using techniques of microfabrication.
Micro motor Gyroscope Accelerometer
Image source: Internet

63. MEMS v/s CMOS
MEMS fabrication is same as IC (CMOS) fabrication, except
• Mechanical Properties
• Feature Size
• Unconventional Materials
• Release Process
Image source: Internet

64. Different Kind of MEMS Sensors
• MEMS Inertial Sensors (Accelerometers and Gyroscopes)
• MEMS Pressure Sensors
• MEMS Gas Sensors
• MEMS Humidity and Temperature Sensors
• MEMS Chemical Sensors
• MEMS Bio sensors

65. MEMS Pressure Sensors
Image source: Internet
Silicon
Substrate
~10um thick
membrane

66. MEMS Pressure Sensors
Image source: Internet
• Pressure results in change in shape
• Change in shape changes resistance
• Change in resistance changes electrical signals

67. Pressure Sensor
Full Bridge Configuration
• Half Bridge and Quarter Bridge configurations also possible
• Sensitivity: Full Bridge > Half Bridge > Quarter Bridge
• What is Sensitivity of a sensor ?
Image source: Internet

68. Micrograph, Sensitivity
Courtesy: Prof K N Bhat, IISc

69. -
+
R2
R2R1
R1
Vout = ?
V-
V+
Flex Your Neurons

70. -
+
R2
R2R1
R1
Vout = ?
V-
V+
Vout = (R2/R1) (V+ - V-)
Flex Your Neurons!

71. -
+
R2
R2R1
R1
Vout
V-
V+
-
+
+
-
Amplifier with Voltage Follower

72. -
+
R2
R2R1
R1
V-
V+
-
+
+
-
Rf
Rf
Rg
Flex Your Neurons!
Vout = ?

73. Vout = (1+2Rf/Rg)(R2/R1) (V+ - V-)
Flex Your Neurons!
-
+
R2
R2R1
R1
V-
V+
-
+
+
-
Rf
Rf
Rg
Vout = ?

74. Types of MEMS Accelerometer
• Capacitive
• Peizoelectric
• Tunneling
• Peizoresistive
• Thermal

75. MEMS Accelerometer
Working principle
Image source: Internet

76. Operating Principle
Image source: Internet

77. Capacitive MEMS Accelerometer
Image source: Internet

78. MEMS Gyroscope
Image source: Internet

79. Types of MEMS Gyroscope
Tuning Fork Gyroscopes
• Rotation causes the proof masses to vibrate out of plane
• The vibration is sensed capacitively with a CMOS circuit
Image source: Internet

80. Vibrating-Wheel Gyroscopes
• Capacitive sensing under the wheel
• Can be used to detect two in-plane rotational axes
Image source: Internet
Types of MEMS Gyroscope

81. ax
az
ay
x0, y0 , z0
x1, y1 , z1
x1 =
2.t
x
a
2
1
.t
x
v
0
x 
y1 =
z1 =
2.t
y
a
2
1
.t
y
v
0
y 
2.t
z
a
2
1
.t
z
v
0
z 
Motion Sensing With Accelerometer
Accelerometer is for linear motion

82. x
y
z
Roll =
Pitch =
Yaw =
Where is the angular rate
readings from gyroscopes
trx
trz
try
r
Motion Sensing With Gyroscope
Roll
Pitch
Yaw
• Gyroscope is for angular motion
• Real objects have finite shape
• Real Motion is a combination of linear and angular motions

83. Current Trends in Sensor Technology
•6-axis IMU
•3-axis Accelerometer
•3-axis Gyroscope
•Motion Processor
•4mm x 4mm
MPU-6050
•9-axis IMU
•3-axis Accelerometer
•3-axis Gyroscope
•3-axis Magnetometer
•Motion Processor
•4mm x 4mm
MPU-9150
•9-axis IMU
•3-axis Accelerometer
•3-axis Gyroscope
•3-axis Magnetometer
•Motion Processor
•3mm x 3mm
MPU-9250
Case: Inertial Measurement Unit (IMU)
Image source: www.invensense.com

84. Invensense MPU-9X50
Image source: www.invensense.com

85. • (Non) Linearity
• The quantity to be sensed
• Ambient conditions
– Temp, Humidity etc
• Operating conditions
– Voltage, Current
• Response time
– How quickly a sensor can report changes
• Hysteresis
– Sensing while ascending vs descending
Sensor Characterization

86. Perfect
Motion
Sensors
Perfect
Motion
Model
Perfect
Path
Estimation
(Motion model not required) (Motion sensors not required)
Working with Low-cost Motion Sensors
Low-cost
Motion
Sensors
Motion
Model
Improved
Path
Estimation
Fusion

87. Example: Pedestrian Navigation
Motion model:
Zero Velocity Update (ZUPT)

88. Summary
• Current trend
• Enabling technology
• MEMS pressure sensor
• MEMS accelerometer and gyroscope
• Motion estimation with accelerometer and gyroscope
• Sensor characterization
• Fusion of low-cost sensors and motion model
• Motion model (ZUPT) for pedestrian navigation

90. Calibration
• Why is calibration required ?
• What is calibration ?
• What are the ways to calibrate inertial sensors ?
• What is the outcome of calibration process ?
• How is calibrating one sensor different from
calibrating sensor array ?

91. Some random errors in sensors
• Randomness in mother nature gets manifested in the
sensors’ structures during fabrication and packaging
• Some errors may come after prolonged use
• Further there are some errors that occur based on the
environment/operating conditions.
These errors are corrected by calibrating the sensors.
Why is Sensor Calibration Required

92. • Variation during fabrication & packaging are random
in nature
• Each sensor from the same lot or same fabrication
house would be different
No two fabricated sensors are same !
Each sensor must be calibrated before use !
Does Each Sensor Require Calibration

93. • Sensor parameters are compared with any standard
reference to find the error.
• The error in any measured parameter can be modelled
as gain and bias
Where k is the gain and b is the bias
• The process of correcting sensors output with gain
and bias known as calibration compensation.
• On the fly calibration-suitable feedback mechanism.
bkpp measuredcalibrated  *
Calibration

94. oblu – A Shoe-mounted Indoor GPS

95. • The MEMS IMUs are placed
in an array of 2x2
• Within every IMU, 3-axis
accelero and 3-axis gyros are
present
• Within IMU, accelerometers
are orthogonal to each other
• Within IMU, gyroscopes are
also orthogonal to each other
• Orientation of all the IMUs is
same
– Accelero in x/y/z direction in
all the IMUs are aligned by
design
– Gyroscopes in x/y/z direction
in all the IMUs are aligned by
design
oblu – A Multi-IMU Platform

96. Randomness during fabrication get embedded as following:
• The axes x-y-z within an IMU may not be orthogonal to
each other
• One IMU’s coordinate axis may not align with rest of the
IMUs
• Other obvious errors may appear in the form of gain/bias
in individual sensors.
All the above non-idealities can be modelled as gains &
biases !
Error in oblu

97. • Accelerometer are calibrated using 3 axis-Shakers Table
• Gyroscopes are calibrated using 3 axis-Rate Table
• Both these instruments cost thousands of dollars.
• A low cost and less time consuming solution is required
Sophisticated Calibration

98. • A known standard reference acceleration is available
everywhere
– Acceleration due to gravity
• z-axis accelerometers will sense “g” when oblu is placed
horizontally.
• x and y axis will also sense g when placed vertically.
• All three axis will measure components of “g” if placed at
certain angle.
Gravity as Standard Reference

99. • A 3D printed icosahedron with cavity to hold oblu
• Compare “g” as obtained from oblu and with the expected
value
• Repeat the process for 20 different angles at which oblu
is placed
Icosahedron: Calibration Device

100. • oblu’s data is collected and compared against the expected
output for 20 different cases
• Only 12 cases are enough for calculating the gain and
biases
Calibrating “oblu”

101. • No low cost standard reference for gyroscopes is
present.
• So only bias (offset) of the gyroscopes is estimated
• At static condition the angular velocity reading is
considered as bias.
The offset error can easily be estimated and eliminated !
Estimating Bias of a Gyroscope

102. • The errors that arise due to noise is eliminated by the Zero
Velocity Update (ZUPT) Algorithm.
• Human stride shows a standstill moment or Zero Velocity
• The standstill phase of stride is called “step”.
• Imperfect sensor measure non-zero velocity at standstill
• Non-zero velocity at standstill is due to noise in the system
• Fine tune the computation model based on this information
On-the-fly Calibration

103. • IoT, Sensors and Calibration - Whats the relation? by
Amit and Subho
– https://www.linkedin.com/pulse/iot-sensors-calibration-whats-
relation-amit-k-gupta
• Story of a Shoe-mounted IoT Sensor Calibration by
Amit and Subho
– https://www.linkedin.com/pulse/story-shoe-mounted-iot-sensor-
calibration-amit-k-gupta
Refrences

104. • Random variations in fabrication process introduce random
errors
• Each sensor has different characteristic due to randomness
• All the errors can be modelled as gain/sensitivity and
bias/offset
• Process of identifying gain & offset is known as calibration
• Each and every sensor must go through calibration
• Sensor’s output is calibration compensated at post
processing
• Calibration needs to be performed periodically
• Non-orthogonality of axis is also a source of error in motion
sensors
• Misalignment of IMUs’ coordinate system is source of error
in multi-IMU system
• Step detection (ZUPT approach) is used for on the fly
calibration and compensation for shoe-mounted PDR sensors
Summary

105. Thank You
Contact Us
hello@oblu.io
R&D Centre: 171 MIG, Awadhpuri, Block B, Lakhanpur, Kanpur, India, PIN – 208024
www.inertialelements.comwww.oblu.io

106. Internal Resistance
Rint
EMF
• OCV or EMF (Li-ion): 2.7 (min), 3.7 (typ), 4.2 (max)
• Rint: ~500 mOhms
• Vbat = EMF – Iload * Rint
Vbat
Iload ≡ (Ityp, Ipeak)