Phillip Stanley-Marbell
Foundations of Embedded Systems
Department of Engineering, University of Cambridge
http://physcomp.eng.cam.ac.uk
Topic 15: Quantifying Power, Energy, Performance, and Noise
(~35 minutes)
Version 0.2020
Pre-Recorded
Video
23
Department Teaching Survey
2
http://to.eng.cam.ac.uk/teaching/surveys/4B25_Mich.html
(link)
23
Intended Learning Outcomes for this Topic
3
Quantify power dissipation, heat, performance, and noise in embedded systems
By the end of this topic, you will be able to:
Identify the key sources of power dissipation in an embedded system
Design more power-, performance-, and error-efficient embedded systems
Recall the analytic relationships governing power and performance in CMOS circuits
23
4
“You can only analyze the data you
have. Be strategic about what to
gather and how to store it.”
— Marie Skłodowska Curie.
23
Nutshell: Measurement of System from Topic 11
5
23
Quantifying Eciency, Part 1: Computation and Sensing
6
BMX055
(accelerometer,
gyroscope, and
magnetometer)
MMA8451Q
(accelerometer)
LPS25H
(pressure)
HDC1000
(humidity)
MAG3110
(magnetometer)
L3GD20H
(gyroscope)
BMP180
(pressure)
TCS3772
(light)
TMP006B
(infrared)
Si7021
(humidity and
temperature)
ADXL362
(accelerometer)
ARM Cortex-M0+
(processor)
Size reference:
2nd Generation
1st Generation
Miniature Warp subset
Warp
Recall, from
Topic 11
Next: Quantifying the
efficiency of this system
42
7
42
8
23
9 9
0.6 grams (batteries included)
Quantifying Efficiency
1 of 3
: Mass
14 mm
10
Next: Quantifying Power Dissipation…
Battery (1 of 2)
Flash Memory
Temperature-compensated realtime clock
Battery (2 of 2)
Temperature, humidity, pressure,
and gas (CO and VOCs) sensor
ARM Cortex-M0+
Connector
23
Quantifying Efficiency
2 of 3
: Power Dissipation
11 11
23
Measuring the Sleep / Awake Power Dissipation
12
23
Measuring the Sleep / Awake Power Dissipation
13
23
Measurements (But Not Yet Quantitative Enough)
14
You should analyze the uncertainty in your
measurements (a relevant example later)…
23
Quantifying Efficiency
3a of 3
: Precision and Uncertainty
15
Sensor Supply Voltage Range Accuracy Range Interface Precision Range
(V) (Noise Measure) (bits/sample)
MMA8451Q accelero meter 1.95 3.6 99 126 µg/
p
Hz 8or14
BMX055 accelerometer 2.4 3.6 150 µg/
p
Hz 8or12
ADXL362 accelerometer 1.6 3.5 175 550 µg/
p
Hz 4, 8, or 12
L3GD20H gyroscope 2.2 3.6 0.011 °/s/
p
Hz 8or16
BMX055 gyroscope 2.4 3.6 0.014 °/s/
p
Hz 8or16
MAG31 10 magnetometer 1.95 3.6 0.25 0.4 µT 8or16
BMX055 magnetometer 2.4 3.6 0.3 1.4 µT 8or13(x-, y-), 15 (z-)
SI7021 hygrometer 1.9 3.6 ±2% accuracy 8, 10, 11, or 12
±0.025–0.2% precision
HDC1000 hygrometer 3.0 5.0 ±4% accuracy 14
±0.1% precision
LPS25H barometer 1.7 3.6 0.01 0.03 hPa 8, 16, or 24
BMP180 barometer 1.6 3.6 0.03 0.06 hPa 8, 16, or 19
HDC1000 thermometer 3.0 5.0 ±0.2°C 14
SI7021 thermom et er 1.9 3.6 ±0.3°C 11, 12, 13, or 14
ADXL362 thermometer 1.6 3.5 ±0.5°C 4or12
TMP006B thermometer 2.2 ±1°C 8or14
BMP180 thermometer 1.6 3.6 ±1°C 8or16
MAG31 10 thermometer 1.95 3.6 ±1° 8
L3GD20H thermometer 2.2 3.6 ±1° 8
LPS25H thermometer 1.7 3.6 ±2°C 8or16
BMX055 thermometer 2.4 3.6 ±2°C 8
TCS3772 photometer 2.7 3.3 14%–35% Irradiance Responsivity 8or16perR/G/B/clear
23
Quantifying Efficiency
3b of 3
: Precision and Uncertainty
16
       






-    ()
 
Mean: 22.6255, Stdev.: 0.302978
Kurtosis: 2.5615 (3 is Gaussian), Skewness: -0.0624255,
Coefficient of Variation of Abs[data]: 0.013391,
Min: 21.9428, Max: 23.2856,
Significant Decimal Digits in Measurement Batch: 1
Significant Binary Digits in Measurement Batch: 3
Est. Dist.: NormalDistribution[22.6311, 0.322718]
The null hypothesis that the data is distributed according
to the NormalDistribution[x
.
,y
.
] is not rejected at the 5.
percent level based on the Cramér-von Mises test.
     






 
 









-    ()
 
Mean: 5.06668, Stdev.: 0.494651
Kurtosis: 89.8772 (3 is Gaussian), Skewness: -9.21422,
Coefficient of Variation of Abs[data]: 0.0976282,
Min: 0.274667, Max: 5.43229,
Significant Decimal Digits in Measurement Batch: 0
Significant Binary Digits in Measurement Batch: 2
Est. Dist.: StudentTDistribution[5.12832, 0.0956067, 2.18457]
The null hypothesis that the data is distributed according
to the NormalDistribution[x
.
,y
.
] is rejected at the 5.
percent level based on the Cramér-von Mises test.
     






 
 
23
Quantifying Eciency, Part 2: Displays
17
Next: Quantifying the
efficiency of this display…
Govisionox Optoelectronics
1.2 inch 390x390 AMOLED Display
23
Measurements: Setup
18
Source/Measure Unit (SMU)
Keithley 2450
Current measurement accuracy 10nA
Govisionox Optoelectronics
1.2 inch 390x390 AMOLED Display
23
19
23
Measurements (But Not Yet Quantitative Enough)
20
You should analyze the uncertainty in your
measurements (showed a relevant example earlier)…
23
Further Reading
21
Best next step:
▶︎ Like learning to swim, you can’t learn all you need from a textbook
▶︎ Quantify the efficiency of the system you implement for your final project
Test your understanding:
▶︎ Complete these online self-assessments on https://f-of-e.org/
https://f-of-e.org/chapter-15/#exercises
23
Things to Do
22
Complete a “muddiest point” 2-question survey using this link
CUED Teaching Survey: http://to.eng.cam.ac.uk/teaching/surveys/4B25_Mich.html
(link)