Master project

OutlineEmbedded digital electronics

1 8-bit microcontroller

2 32-bit microcontroller (bare metal)

3 32-bit microcontroller (libraries)

4 32-bit microcontroller (executiveenvironment)

5 32/64-bit microcontroller (operatingsystem)

6 32/64-bit microcontroller (kernelmodule)

7 32/64-bit microcontroller(kernel-userspace communication)

8 32/64-bit microcontroller/FPGA(Redpitaya co-design)

9 32/64-bit microcontroller/FPGAradiofrequency signal processing

Bottom-up development

Project

1 Raspberry Pi 4 single boardcomputer (4×1.5 GHz CPUs)

2 Python/GNU Radio userspaceapplications

3 Communication with personalcomputer (0MQ)

4 Web-controlled instrument (REMI)

5 Characterization of spectral

properties of radiofrequency device

• frequency sweep networkanalyzer

• broadband noise source +FFT

• pulsed excitation + FFT

6 ringdown measurement of resonantsystem : Q & fr

Top-down development

1 / 16

Master project

Acoustic transducer• Electromagnetic wavelength : λ = 300/fMHz

• At 1 GHz : 30 cm wavelength ⇒ signal processing (e.g. cavities, transformer,filters, waveguides) will be ' λ wavelength

• Acoustic waves propagate ∈ [500..10000] m/s : at 3000 m/s, 105 slower

• Convert between electromagnetic and acoustic wave using piezoelectricsubstrates : 30 cm become 3 µm wavelength

• 40-50 filters/phone, 1.4 billion phones/year ⇒ >50 billion acoustic filters/year 1

8/24/2016 10:22:53 AM1311.6010K42-103899-La

Trc1 S21 dB Mag 10 dB/ Ref -50 dB Cal 1

Ch1 Start 100 MHz Pwr 0 dBm Bw 10 kHz Stop 200 MHz

-90

-80

-70

-60

-50

-40

-30

-20

-10

-100

0

M1

M2

-50 dB

M1 134.3981 MHz -60.8503 dBM2 137.4600 MHz -9.9392 dB

Trc2 S21 Phase 50°/ Ref 0° Cal 2

Ch1 Start 100 MHz Pwr 0 dBm Bw 10 kHz Stop 200 MHz

-200

-150

-100

-50

0

50

100

150

200

-250

250

M1

M2

0°

M1 134.3981 MHz -48.78 °M2 137.4600 MHz -121.69 °

1. www.canalys.com/newsroom/canalys-global-smartphone-market-q4-2019 2 / 16

Master project

Acoustic transducer• Electromagnetic wavelength : λ = 300/fMHz

• At 1 GHz : 30 cm wavelength ⇒ signal processing (e.g. cavities, transformer,filters, waveguides) will be ' λ wavelength

• Acoustic waves propagate ∈ [500..10000] m/s : at 3000 m/s, 105 slower

• Convert between electromagnetic and acoustic wave using piezoelectricsubstrates : 30 cm become 3 µm wavelength

• 40-50 filters/phone, 1.4 billion phones/year ⇒ >50 billion acoustic filters/year 1

8/24/2016 10:22:53 AM1311.6010K42-103899-La

Trc1 S21 dB Mag 10 dB/ Ref -50 dB Cal 1

Ch1 Start 100 MHz Pwr 0 dBm Bw 10 kHz Stop 200 MHz

-90

-80

-70

-60

-50

-40

-30

-20

-10

-100

0

M1

M2

-50 dB

M1 134.3981 MHz -60.8503 dBM2 137.4600 MHz -9.9392 dB

Trc2 S21 Phase 50°/ Ref 0° Cal 2

Ch1 Start 100 MHz Pwr 0 dBm Bw 10 kHz Stop 200 MHz

-200

-150

-100

-50

0

50

100

150

200

-250

250

M1

M2

0°

M1 134.3981 MHz -48.78 °M2 137.4600 MHz -121.69 °

1. www.canalys.com/newsroom/canalys-global-smartphone-market-q4-2019 3 / 16

Master project

Acoustic resonator• Dual surface acoustic wave (SAW) resonator operating in the

433.05–434.79 MHz ISM band• Electromagnetic wave : 70 cm wavelength → acoustic wave : 7 µm• 2-port device used either in reflexion or transmission• Q ' 104 @ 434 MHz

8/27/2020 12:33:10 PM1311.6010K42-103899-La

Trc3 S21 dB Mag 2 dB/ Ref -10 dB Cal 1

Ch1 Center 434 MHz Pwr -10 dBm Bw 10 kHz Span 2.5 MHz

-18

-16

-14

-12

-10

-8

-6

-4

-2

-20

0

-10 dB

Trc2 S21 Phase 10°/ Ref -10° Cal 2

Ch1 Center 434 MHz Pwr -10 dBm Bw 10 kHz Span 2.5 MHz

-50

-40

-30

-20

-10

0

10

20

30

-60

40

-10°

SENSeOR SEAS10 sensor

Y (

mm

)

X (mm)

432.443174 MHz

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.5 1 1.5 2

16

18

20

22

24

26

28

30

Y (

mm

)

X (mm)

433.218974 MHz

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.5 1 1.5 2

16

18

20

22

24

26

28

30

Y (

mm

)

X (mm)

434.368974 MHz

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.5 1 1.5 2

16

18

20

22

24

26

28

30

Y (

mm

)

X (mm)

reflectivity

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.5 1 1.5 2

0.006

0.008

0.01

0.012

0.014

0.016

0.018

0.02

4 / 16

Master project

Network analyzerFrequency dependent response of Device Under Test (DUT)

• Network analyzer principle :• sweep continuous wave single tone signal source and observe

transmitted power for each spectral component ⇒ frequency sweepnetwork analyzer (scalar or vector when measuring phase)

• broadband source and analyze each spectral component transmittedby the device (Fourier transform)

• source, shaping, DUT, signal processing, communication and display

• analog network analyzers have existed : noise source and filter array

5 / 16

Master project

Digital electronics 2 3

Why digital ?• stability (algorithms do not drift over time)• tunability (adapt algorithm to measurement conditions, tune

parameters using software tuning only)• flexibility (a single circuit for multiple applications)

V2 →(analog)

←Saturn1(digital)

2. D.A. Mindell, Digital Apollo : Human and Machine in Spaceflight, MIT Press(2011)

3. D.A. Mindell, Between Human and Machine : Feedback, Control, and Computingbefore Cybernetics, Johns Hopkins University Press (2002)

6 / 16

Master project

Principle

Objective : characterize frequency response of a radiofrequency deviceunder test (DUT)

RPi4DVB−T USB

sourceDUT

Eth PC0MQ

GNU Radio

• Instrument control : GNU/Linux running on Raspberry Pi 4 (RPi4)

• Source : broadband source (noise/pulse)

• Acquisition & sampling : R820T2 based DVB-T receiver as generalpurpose Software Defined Radio (SDR) receiver

• Processing : GNU/Radio communicating with PC

7 / 16

Master project

Noise source : reflection mode• Broadband noise source (reverse-polarized Zener diode + amplifiers)• Sample 2.2 MHz-wide bandwidth• Transpose from 434 MHz to baseband

SENSeOR SEAS10 passive wireless sensor : two resoances in the433.05–434.79 MHz ISM band 8 / 16

Master project

Noise source : transmission modeRather than voltage divider bridge to ground, resonator pair in serieswith signal source

Options

Title: Not titled yet

Output Language: Python

Generate Options: QT GUI

Variable

Id: samp_rate

Value: 2.2M

outcommand

osmocom Source

Sync: Unknown PPS

Number Channels: 1

Sample Rate (sps): 2.2M

Ch0: Frequency (Hz): 434M

Ch0: Frequency Correction (ppm): 0

Ch0: DC Offset Mode: 0

Ch0: IQ Balance Mode: 0

Ch0: Gain Mode: False

Ch0: RF Gain (dB): 44

Ch0: IF Gain (dB): 20

Ch0: BB Gain (dB): 20

freq

in

freq

bw

QT GUI Frequency Sink

FFT Size: 4.096k

Center Frequency (Hz): 0

Bandwidth (Hz): 2.2M

in

QT GUI Time Sink

Number of Points: 1.024k

Sample Rate: 2.2M

Autoscale: No

9 / 16

Master project

Pulsed source

0.5 ns Transmission spectra as afunction of pulse duration

0.625 ns

1.0 ns

1.25 ns

+−

threshold

ADCMP580

2.22 ns10 / 16

Master project

Running on the Raspberry Pi4Characterization of a band pass filter using micro-HDMI output of RPi4

cadencement du flux

filtre à caractériser

Options

Title: Not titled yet

Output Language: Python

Generate Options: QT GUI

Variable

Id: samp_rate

Value: 32k

out

Noise Source

Noise Type: Gaussian

Amplitude: 1

Seed: 0

outinThrottle

Sample Rate: 32koutin

Low Pass Filter

Decimation: 1

Gain: 1

Sample Rate: 32k

Cutoff Freq: 6.4k

Transition Width: 3.2k

Window: Hamming

Beta: 6.76

freq

in

freq

bw

QT GUI Frequency Sink

FFT Size: 1.024k

Center Frequency (Hz): 0

Bandwidth (Hz): 32k

#!/ usr/bin/env python3from PyQt5 import Qtfrom gnuradio import qtguifrom gnuradio.filter import firdes

class rpi(gr.top_block , Qt.QWidget):

def __init__(self):gr.top_block.__init__(self , "Title")Qt.QWidget.__init__(self)

self._qtgui_freq_sink_x_0_win = sip.→↪→wrapinstance(self.→↪→qtgui_freq_sink_x_0.pyqwidget (), →↪→Qt.QWidget)

self.low_pass_filter_0 = filter.→↪→fir_filter_ccf (1, firdes.low_pass→↪→( 1, samp_rate , samp_rate/5, →↪→samp_rate /10, firdes.WIN_HAMMING ,→

↪→ 6.76))self.blocks_throttle_0 = blocks.throttle(→

↪→gr.sizeof_gr_complex *1, samp_rate→↪→,True)

self.analog_noise_source_x_0 = analog.→↪→noise_source_c(analog.GR_GAUSSIAN→↪→, 1, 0)

# ######################################## Connections# #######################################self.connect ((self.→

↪→analog_noise_source_x_0 , 0), (→↪→self.blocks_throttle_0 , 0))

self.connect ((self.blocks_throttle_0 , 0)→↪→, (self.low_pass_filter_0 , 0))

self.connect ((self.low_pass_filter_0 , 0)→↪→, (self.qtgui_freq_sink_x_0 , 0)→↪→)

[...]

11 / 16

Master project

Beyond RF device characterization

Autonomous ACARS 4 (plane-to-ground communication) receiverrunning on RPi4

4. https://sourceforge.net/projects/gr-acars/12 / 16

Master project

Useful tools• Buildroot for developing on the RPi4 or qemu

• GNU Radio running on RPi4

• ZeroMQ or UDP socket to stream from RPi4 to PC

• Python TCP server on RPi4 to receive commands from PC (telnetor Python client)

• Include TCP server in GNU Radio (Python Module)

• Launch TCP server from GNU Radio (Python Snippet) with accessto the callback functions

• Display on PC stacked spectra with broader bandwidth than theDVB-T 2.8 MHz bandwidth

13 / 16

Master project

Preliminary investigationsWhy a pulse and what properties are wanted ?

1 What is the spectrum of a Dirac function ?2 What is the spectrum of a square function with period T ?3 What is the spectrum of a pulse with rise time τ and duration T ?4 What is the spectrum of a pulse with duration T gating a sine wave

with frequency f , f � 1/T ?

0

0.2

0.4

0.6

0.8

1

-1 -0.5 0 0.5 1

norm

aliz

ed s

pectr

um

(a.u

.)

normalized frequency (a.u.)

2/1024

64/1024

512/1024

16/32

sin/1024

x=z e r o s ( 1 0 2 4 , 1 ) ; x ( 1 : 2 ) =1;x=[x ; x ; ; x ; ; x ; x ] ;x=x−mean ( x ) ;f=l i n s p a c e (−1 ,1 , l e n g t h ( x ) ) ;y= f f t s h i f t ( abs ( f f t ( x ) ) ) ;p l o t ( f , y /max ( y ) )

x=z e r o s ( 1 0 2 4 , 1 ) ; x ( 1 : 6 4 ) =1;x=[x ; x ; ; x ; ; x ; x ] ;x=x−mean ( x ) ;y= f f t s h i f t ( abs ( f f t ( x ) ) ) ;p l o t ( f , y /max ( y ) )

x=z e r o s ( 1 0 2 4 , 1 ) ; x ( 1 : 5 1 2 ) =1;x=[x ; x ; ; x ; ; x ; x ] ;x=x−mean ( x ) ;y= f f t s h i f t ( abs ( f f t ( x ) ) ) ;p l o t ( f , y /max ( y ) )

x=z e r o s ( 1 0 2 4 , 1 ) ;f o r k =1:1024

i f mod( k , 1 6 )<8 x ( k ) =1; endendx=[x ; x ; ; x ; ; x ; x ] ;x=x−mean ( x ) ;y= f f t s h i f t ( abs ( f f t ( x ) ) ) ;p l o t ( f , y /max ( y ) )

x=s i n ( [ 0 : 1 0 2 3 ] / 1 0 2 4∗2∗ p i ) ’ ;x=[x ; x ; ; x ; ; x ; x ] ;y= f f t s h i f t ( abs ( f f t ( x ) ) ) ;

14 / 16

Master project

Conclusion & perspectives

20 GNU/Linux Magazine France N°240 https://www.gnulinuxmag.com

MOTS-CLÉS : RADAR, GNU/OCTAVE, TRAITEMENT DU SIGNAL

ANALYSE ET RÉALISATION D’UN RADAR À BRUIT PAR RADIO LOGICIELLE

J.-M FRIEDT[FEMTO-ST, département temps-fréquence, Besançon, France]

W. FENG[Xidian University, National Laboratory of Radar Signal Processing, Xi’an, Chine]

IA, ROBOTIQUE & SCIENCE SDR

Le traitement du signal, domaine abstrait par excellence, devient très amusant lors de sa mise en pratique. Nous nous proposons de démontrer expérimentalement un RADAR à bruit mis en œuvre par radio logicielle.

Pour paraphraser l’introduction d’une vidéo de chanteurs de rap français [1], « on a 3 mi-nutes, un appart’ rempli de

bordel, une vieille émission pour les gosses dans l’ordi, on peut sûrement y aller au bluff ». Dans notre cas, nous n’avons pas 3 minutes, mais 5 se-maines de confinement, on a bien un appartement rempli de bordel, et plu-tôt que l’émission de télévision, nous avons des antennes et des circuits de radio logicielle. Pouvons-nous mesurer la distance de la maison en face dudit appartement en réalisant un RADAR (RAdio Detection And Ranging) actif avec les moyens du bord (figure 1) ?

NOTE

Cet article est le premier d’une série de trois articles sur le sujet.

FORK ME !

• Opportunity to deploy anembedded system for apractical application

• Opportunity to learn howto properly use GNU/Linuxtargeted to an embeddedenvironment (Buildroot –never compile on thetarget system)

• Reproducible experiment :<50 euros (RPi4+DVB-T)

• Radiofrequency signalprocessing ...

• for characterizing a (RF)microsystem.

• RADAR range resolution :∆R = c/(2B) ...

• ... can be reached withmany sources other than apulse generator to achievelarge B.

15 / 16

Master project

References

• C. Simmonds, Mastering Embedded Linux Programming, PacktPublishing (2015)

• K. Yaghmour & al., Building Embedded Linux Systems : Concepts,Techniques, Tricks, and Traps, O’Reilly (2008)

• P. Ficheux & E. Benard, Linux embarque, Eyrolles (2012)

• P. Ficheux, Introduction a Buildroot, GNU/Linux Magazine France(Avril 2010)

• M. Corbin, Buildroot for RISC-V, FOSDEM 2019 a https://

archive.fosdem.org/2019/schedule/event/riscvbuildroot/

• Depot Buildroot a https://github.com/buildroot/buildroot

• Experimenter ...

16 / 16