Home

Measurement of changing mechanical properties of carbon

1. Command which creates a task is called zTaskCreate and it has six parameters xTaskCreate pvTaskCode pcName usStackDepth pvParameters uxPriority pxCreatedTask Each parameter is shortly described in the table Tab 6 1 All created tasks are started by scheduler s start command vTaskStartScheduler With this command all tasks start doing their functions All running tasks has their own part of RAM memory space which in sum must be lower then dedicated size of memory for FreeRTOS This prevents any dangerous state or data loss 6 COMMUNICATION WITH OTHER BOARDS o Tab 6 1 xTaskCreate Command Parameters 22 Parameter Description puTaskCode Function pointer pcName Task name usStackDepth Stack size puParameters Parameters of the task uxPriority Task priority pxCreated Task Used for backward handle Every task usually works in infinite loop similar to main function The next Code 6 1 represents example of a task without any function only body 22 Code 6 1 Main loop of any task int task_name void pvParameters initialization here while 1 do something here return 0 6 4 Communication with HM Board I would like to express big thanks to my colleague ing Tom s B ca for help with the basics of implementing FreeRTOS and CSP to the HM board Asis written previously all boards communicate via CSP and the HM board has several commands coupled wit
2. surement of mechanical changes of car bon fibre material The board is one of the part nanosatellite VZLUsat 1 and one of experiments on board Micro controller processing sampling of signal calculates Fast Fourier Transform and attenuation of signal All measurements of the probe VZLUsat 1 will be launched during mission QB50 KEYWORDS QB50 CubeSat Space FFT Compos ite Carbon VZLUSAT1 Sampling Re search ANOTACE Tato diplomov pr ce pojedn v o m en mechanick ch vlastnost uhl kov ho kompozitu M c deska je jednou z mnoha dal ch na nanosatelitu VZLUsat 1 a jednou z experiment Mikrokontrol r zpracov v navzorko van sign l spo t rychlou Fourierovu transformaci a tlum sign lu Ve ker m en na sond VZLUsat 1 budou vypu t na b hem mise QB50 KL OV SLOVA QB50 CubeSat Vesm r FFT Kom pozit Carbon VZLUSAT1 Vzorkov n V zkum NENTVICH Ond ej Measurement of changing mechanical properties of carbon com posite on nanosatellite miniCube mission QB50 master s thesis Prague Czech Tech nical University in Prague Faculty of Electrical Engineering Department of Microelec tronics 2015 75 p Supervised by Ing Ladislav Sieger CSc ACKNOWLEDGEMENT would like to express thanks to the mentor Mr Ing Ladislav Sieger CSc for many useful advices comments and patience during works on nanosatellite this master s thesis and also to my col
3. Drawings of HM panel 4 8 1 Young s Modulus of Elasticity Young s modulus is synonym for elastic modulus It measures stiffness of an elastic material to characterize materials Modulus is characterized as ratio of stress to strain and used for calculation of natural freguencies of the cantilever 11 The issue about natural freguencies is described in detail in the thesis of my colleague Martin Urban 1 W 4 8 2 Attenuation of Signal Measuring attenuation of the signal is made for verification of elasticity of mate rial This parameter is only additional to natural frequencies In the picture Fig 4 6 is sample of signal which has been measured on the HM panel It produced damped oscillations with almost exponential envelope There are several ways how to get exponential envelope One is the electrical way that rectifies the signal and filters it with low pass filter There is one disadvan tage though because of the low amplitude only single diode rectifier can be used Another option for this purpose is to design a rectifier made of operation amplifiers This way has been rejected because of higher consumption of energy and it is not necessary to implement it in a physical way 4 MECHANICAL CHANGES MEASUREMENT o 400 01 02 03 04 05 06 07 08 09 1 t s Fig 4 6 Sample of signal 4 9 Computing Process of FFT Fast Fourier Transform is based on Discrete Fourier Transform DFT equation 4
4. Optics and Measurement Conference 2014 2015 no 1 DOI 10 1117 12 2175925 11 no Elastic Modulus online 2015 5 18 cited 2015 06 23 Available from https en wikipedia org wiki Elastic_modulus 59 75 12 DUNN Patrick F Measurement and data analysis for engineering and science 2nd ed CRC Press Taylor amp Francis 2010 490 p ISBN 978 1 4398 2568 6 13 Rychl Fourierova transformace FFT pro AVR online 2012 02 28 cited 2014 07 14 Available from http elektronika kvalitne cz ATMEL necoteorie transformation AVRFFT AVRFFT html 14 Signal Flow Graphs of Cooley Tukey FFTs online Available from http cnx org contents e460644d c1d6 4dee a60e 3ee5220e88ba0 11 Appendix_1 _FFT_Flowgraphs 15 8 16 bit Atmel XMEGA Microcontroller ATrmega128A4U ATxmega64A4U ATrmega32A4U ATrmega16A4U online 2014 03 cited 2014 07 17 Available from http www atmel com images atmel 8387 8 and16 bit avr microcontroller xmega a4u datasheet pdf 16 Oscil tory TCXO a VCTCXO online cited 2015 04 19 Available from http www krystaly cz cs Produkty Oscilatory TCX0 17 23LCV1024 online 2012 09 cited 2015 01 24 Available from http ww1 microchip com downloads en DeviceDoc 25156A pdf 18 Thermometer ADT7420 online 2012 12 cited 2014 09 13 Available from http www analog com media en technical documentation data sheets ADT7420 pdf 19 SIGNAL CHAIN BASICS Part 32 Digital
5. Tab 5 1 Parameters of MCU ATxMegal28A4U 15 Type Parameters Description Flash 128 kB Size of internal program memory SRAM 8kB Size of internal data memory EEPROM 2kB Size of internal data memory non volatile Speed up to 32MHz Maximal speed of external internal oscillator crystal ace ADC up to 12 Number of single ended inputs ADC up to 8 Number of differential inputs ADC 12 bits Resolution of ADC Timers Counters 5 Number of 16 bits Timers Counters in MCU Supply 2 7 3 6 V Tolerance of supply voltage in range of fre quencies 0 32 MHz E 5 2 2 Oscillator The source of accurate clocks is necessary for signal sampling It means finding the best crystal or oscillator for this purpose Common crystals have low tempera ture stability about 25 ppm in 30 C to 80 C temperature range The oscillator TCX0 1A is very precise and is tuned to 16 470 MHz Frequency stability vs temperature is better then 2ppm Oscillator usually comes with an internal capacitive trimmer but for space mission it would be very unsuitable because it could change freguency during the mission and the results would be invalidated Evaluation depends mostly on freguency change of cantilever in time That capacitive trimmer was replaced by solid capacitor with fixed value This was special reguirement for an oscillator 16 28 75 o 5 2 HM board m 5 2 3 External Memory For high number of points of FFT
6. or write low Each byte is ended by ACK After address could follow data byte s from master slave it depends on read write bit When transmission is ending master sends NACK bit and then stop condition follows SCL rising edge prevents SDA rising edge Data can change only when SCL is low 19 20 7 Bit CA 8 Bit 8 Bit ACK SDA ADDRESS Ci ACK DATA ACK DATA SCL 1 7 8 9 9 1 8 9 5 START STOP Condition Condition 1 8 Fig 6 2 Chart with communication via I C 19 6 2 CubeSat Space Protocol CubeSat space protocol is unified for small nanosatellites such as are in mission QB50 The protocol is similar to TCP IP protocol with simplified header Header has 32 bits and contains following parts image representing it is in Fig 6 3 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1lo F H X Prio R MIT R C y Source Destination Destination Port Source Port Reserved rity AJAJE D R 6 Fig 6 3 CSP header Priority is when master has more then one packet to send packet with higher priority will be send sooner Source is address of master from whom packet came Destination is address where the message will be sent to Destination Port is internal port of slave After receiving a packet the board performes a specific task assigned to that port Source Port is internal po
7. 29 75 5 DEVICE FOR MECHANICAL CHANGES MEASUREMENT o o for catching peaks of voltage from the coil If the diode would be omitted it could cause damage the transistor in the best case In the worst case it could destroy whole satellite So it is necessary The second switch is used for separation of 5 V and 3 3 V power supply 5 V to the gate of transistor is more suitable for high current through power transistor T1 Energy for high power peak is stored in two tantalum capacitors with total ca pacitance of approx 3 mF It is enough for excitation of the coil with high amplitude on the piezo Amount of energy can be controlled by width of pulse to the Piezo on E 5 2 5 Piezo Connection Piezo element is connected to the MCU s ADC through resistor divider Schematic of input measurement is in the picture Fig 5 5 R7 X2 2 RIEZO V PIEZO Piezo CI 10k X2 3 lt 3 oo o X2 4 ae Shielding X2 1 53261 04 AGND AGND Fig 5 5 Schematic of piezo connection Divider reduces amplitude of the input signal Signal is alternating around zero voltage and minimum amplitude of input voltage must be greater then 0 4V otherwise signal is clipped by internal diode Unfortunately this problem was dis covered in time when board was already created and assembled and there was not enough time to change it In the end it was not a problem measured signal could be trimmed slightly by width of pulse to the coil and t
8. 6 4 1 Measure Starts 37 6 5 6 4 2 Returns Signal to DK 6 4 3 Returns Results of Measurement 37 6 4 4 Returns Temperature of HM Board 37 Data Keeper 38 7 Measurement Process in HM board 39 7 1 Read Parameters of Measurement 39 7 2 Get Temperature Orientation and Time 40 7 3 Excite of Coil and Sampling of Signal 41 7 4 Signal Processing FFT 41 7 4 1 Decimation and Address Bit Reversing 41 7 4 2 FFT Process 43 7 5 Signal Processing Attenuation 44 7 6 Store Results 45 1 7 Conclusions eres 45 8 Measurements Testing and Results 47 8 1 Beginning of Research 48 8 2 Research During Internship in Japan 49 8 3 Testing of Pulse Width 50 8 4 Different Climate Conditions 51 8 5 Final Implementation 54 8 6 Final Measurements Before Flight 54 8 7 Conclusion 56 9 Conclusion 57 XII References List of appendices A Schematic of the board B HM board C Article from Japan D Content of DVD 59 63 65 67 69 75 List of Figures 3 1 4 1 4 2 4 3 4 4 4 5 4 6 4 7 4 8 4 9 4 10 5 1 5 2 5 3 5 4 9 9 6 1 6 2 6 3 8 1 8 2 8 3 8 4 Nanosatellite VZLUsat 1 Picture of elementary damped oscillator Simulated damped signal Length of string L vs wavelength A Picture of the Cantilever Health Monitoring HM panel Drawings of Sample of signal Flowchart of computing FFT w
9. Bc Ondrej Nentvich Ing Ladislav Sieger CSc and Kazuo Yana Ph D Czech Technical University in Prague Prague Czech Republic 2 Hosei University Koganei City Tokyo Japan Abstract This paper describes the outcome of internship at the faculty of science and engineering Hosei University in summer 2014 The goal of the project is to design a measuring system of aging properties of a carbon fiber reinforced composite in space The project is a part of the nano satellite project at Czech Technical University in Prague scheduled to be launched in 2016 The measurement environment in space is different from the standard measurements performed on the ground in laboratory The system design specification has a large constraint in size weight and power consumption by the limit of space probes To meet these reguirement the basic measuring system of the mechanical damping characteristics of the carbon fiber composite is designed in this internship project A damping oscillator to simulate the response of the target material has been assembled and measuring parameters are optimized The optimized algorithm has been implemented in the chip to be launched on the space orbit I PROJECT OB50 AND PROBE VZLUSATI During our internship at Hosei University we were working on CubeSat project OB50 concretely on the probe VZLUSATI This project runs under the auspices of Czech Aerospace Research and Test Establishment VZLU and Czech Technical Un
10. CONCLUSION o which could cause the error From this point of view it is required to implement detection to prevent this state use Watchdog or another way to ensure the stability of operation For a space mission it is necessary to implement any system to prevent every hazardous state When any system is not programmed and if board will not respond it will be disconnected from power source and then connected again After that board should work normally Some of the reasons why board could not work properly include a damaged chip memory losing connection with OBC unplugged piezo or coil and also damaged wires When this state happen it is recommended to change plans of reservation time and power for the experiment to zero This could save a little power for running experiments or for what else is needed During this state it is possible to test the communication from time to time and to determine if the results are correct If this is true the plans can change and again include the board which previously did not respond properly References 1 URBAN Martin Measurement of evaporation and evaluation of changes of the mechanical properties of carbon composite on nanosatellite miniCube mission QB50 master s thesis Prague Czech Technical University in Prague Faculty of Electrical Engineering Department of Microelectronics 2015 35 p Super vised by Ing Ladislav Sieger CSc VERONIKA Stehl kov Radiation resistance measurem
11. Czech Technical University in Prague University of West Bohemia Rigaku Innovative Technologies Europe TTS 5M IST Probe VZLUsat 1 consists of many experiments as is shown in Fig 3 1b One of them is X Ray camera called Medipix with Lobster Eye X Ray optics with focal length approx 20cm It takes pictures of the Sun and measures X ray inten sity in range 5 20 keV The important thing is measuring of temperature Satellite 5 75 3 CUBESAT VZLUSAT 1 o o have many PT1000 temperature sensors connected to the one common board called Measure Board Another measurement is HM system which consists of measuring mechanical changes in time radiation shielding and evaporation from the carbon fibre on the probe The goal is to verify properties of carbon fibre material in space Dimensions of the probe in packed state are 20 x 10x 10 cm and when VZLUsat 1 will be dropped out from launcher with other satellites then it deploys solar panels Lobster Eye optics and HM panel Dimensions will change to approx 30x10x10 cm Unpacked state is in the picture Fig 3 1a 3 1 Parts of Probe Each member of the team is responsible for specific part of scientific experiment on board The most important parts are on the followings pages Almost all boards are connected through main 80 pins connector which contains power reserved data signals and also user defined pins which could be used for any purpose 3 1 1 X Ray Optics and Medipix Pr
12. What is a reason why we used decimation from 8 kHz to 500 Hz The result of resolution is sufficient for us up to 200 Hz and sample frequency 500 Hz passed Shannon Nyquist sample theorem with enough reserve This theorem says minimal sampling frequency must be two times higher than maximal measured sampled frequency 40 rectangle window hamming window window blackman window Fig 5 FFT results with window length 4096 samples and different types of windows Results for FFT with signal sampling 500 Hz and window length 1024 are better than previous one resolution is 0 5 Hz but it is still not enough Required is approx 0 1 Hz This limit of calculation allowed us to save memory and to use larger window We made interpolation with help of adding zeros behind the signal into expanded window with window length 4096 points Resolution of the decimated signal to 500 Hz with these calculation parameters is approx 0 12 Hz according to formula 1 Implemented measurement in the program aboard the probe must sample signal just only once so it leads to choose higher sampling rate and then apply decimation on signal to requested sampling rate All these computations were simulated in MATLAB Af IV CALCULATION PROCESS Main process consists of several functions Here is a brief list Sampling and storing data The analog to digital converter included in microcontroller c
13. case that signal is a line in the whole length Although this claim is questionable For safer and more reliable result in autonomous mode it is recommended to use small part of the signal The best part is the beginning In the picture Fig 4 10 is a result of least square The blue line is an outcome directive of linear part of signal The approximation is used in length approx 0 2s Length of least square method is configurable for further improvement of measurement on orbit in depends on quality of sampled signal 5 Device for Mechanical Changes Measurement The idea of measuring mechanical changes is to measure resonance frequency of the material and recursively calculate elastic modulus and other properties So it leads to creation of the measurement system HM panel and HM board 5 1 HM panel Fig 5 1 Picture of HM panel Health Monitoring panel is made from carbon fibre material and milled cantilever shown on the drawing Fig 4 5 HM panel is mounted as tilting panel on the top of the satellite as shown on Fig 3 1a The HM panel consists of some important parts for measuring like excitation coil which attract permaloy target glued on the cantilever under the coil It causes 25 75 5 DEVICE FOR MECHANICAL CHANGES MEASUREMENT a o o oscillations which are measured by piezo electric element glued on the most stressed position at the opposite side of the cantilever Piezo is situated in place of the h
14. coil which attracts cantilever with glued permaloy circle It causes vibrations of the beam and produce mechanicaly damped oscillations which are measured by piezo electric element Piezo transforms mechanical oscillations into electrical Then Microcontroller samples them and evaluate them Process of evaluation consists of calculation of Fast Fourier Transform FFT and attenuation of the signal Results of the measurement are frequencies one is mechanical resonance of the beam the other frequencies are resonances of whole satellite and the last measured parameter is attenuation of beam for additionally specifying of the model using Finite element method FEM Measurement board runs under Free Real Time Operating System FreeRTOS and communicates with the probe through CubeSat Space Protocol CSP via Inter Integrated Circuit I C interface 1 75 2 Mission QB50 Main idea of the mission QB50 is that anybody could build their own probe especially Universities with quite low cost start and required equipment Basic cost for Hardware HW is in range 50 100 thousands Euro Name 38950 is derived from number of CubeSats Generally 50 CubeSats are carried out on orbit These probes are from whole the world One of them is Czech nanosatellite VZLUsat 1 which is the first probe that will be launched during mission QB50 Each satellite is normalized to units which must be observed Basic idea of the mission QB50 is discovering the least exp
15. copy_signal_attenuation copy and decimate signal average_signal rectify average and logarithm the signal for i 0FFSET AVERAGING POINTS i lt int OFFSET_AVERAGING_POINTS n i loaded value spi mem read word i 2 MEM ATTENUATION BEGIN point float loaded value sum x i y i point sum x i i sum y i loaded value sum x i sguared i i recalculation from samples to the time sum x i i fs sum x i fs sum x i sguared fs fs reg k n sum x i y i sum x i sum y i n sum x i sauared Sum x i sum x i 100 reg g sum x i sguared sum y i sum x i sum x iy 3 n sum x i sguared sum x i sum x i where OFFSET AVERAGING POINTS is constant offset in time samples since least square is computed MEM ATTENUATION BEGIN is offset in memory where stored sig nal for computing attenuation begins n is number of averaged points The function copy signal attenuation is only copying signal from one part of external memory to the other and decimate the signal according to DECIMATION_ FACTOR 44 75 o 7 7 Conclusion The function average signal consists of the following processes rectifying of signal using abs function make absolute value moving average for curve smoothing and logarithm them for linearisation of signal according to chapter 4 10 1 and 4 10 2 Code has one main for cycle where all sums in this loop according to equation 4 27 T
16. interfaces con t The I2C Bus online 2009 06 08 cited 2015 06 19 Available from http www planetanalog com document asp doc_id 527900 20 I C Interface online cited 2015 02 13 Available from http www i2c bus org 21 Resources for CSP online 2013 cited 2015 02 13 Available from https github com GomSpace libcsp 22 FreeRTOS online cited 2015 04 09 Available from http www freertos org 23 UFFS online cited 2015 06 26 Available from https sites google com site gouffs 60 75 24 Development kit for XMEGA A4 microprocessors online 2015 cited 2015 07 13 Available from http www eeas cz q en node 68 q 25 Kazuo Yana Ph D Consultations Summer 2014 Hosei University Tokyo Japan 26 URBAN M NENTVICH O STEHL KOV V SIEGER L YANA K Measuring carbon fiber aging on orbit Hosei science collection report 2014 vol 26 no 50 p 1 6 ISSN 2188 8507 List of appendices A Schematic of the board B HM board C Article from Japan D Content of DVD 65 67 69 75 T T nesus 22150 E snad 348GUNN Yusunsog GaTroued UH JUIL 9ued WH aNov aNov o E RD a ale als rx ans ans i OL T r Homs anz gt hee A caida ie a a p NS po ms i sulod set 0838 03 026144 ano ZHINOLY 9L WL OXO L NO 92213 ad oax 700
17. is necessary to have more memory space Internal RAM has only 8kB and almost 3 4 of is used for variables and FreeR TOS Because of that external memory connected to the MCU is necessary FRAM memory is valuable for use in space because it is more reliable then EEPROM and SRAM But it is quite difficult to find a suitable chip All commonly have parallel interface so it is necessary to have many Input Output I O pins for connection to the memory EEPROM has one disadvantage and that erase and write cycle takes a lot of time about 5ms So it led to a choice of SRAM memory It has al most instant write and read cycle but when power supply is disconnected data are lost For computing FFT data loss does not matter required results are stored in data keeper As external memory has been chosen 1 Mb of SRAM Serial Peripheral Interface SPI memory 23LCV1024 17 B 524 Power Switch Power switch on the board serves for excitation of coil which attracts permaloy on the cantilever Detailed schematic is in the following picture Fig 5 4 53261 04 COIL X1 1 D2 MURDE20CTGH CO x1 4 Coil 3 3V 33V 6 R5 o x E desk L 5 je 102 EN EN dida m S D 6 4 1 E BSS123 GND GND GND GND PIEZO ON Fig 5 4 Schematic of power switch Power switch consists of parts such as power MOSFET connected as switch There is a reversed polarized diode between drain of the transistor and power source
18. leagues who support me Also would like to express huge grat itude to Mr Kazuo Yana Ph D from Hosei University in Tokyo Japan who provided me experiences about signal processing during summer internship in 2014 DECLARATION declare that have written my mas ter s thesis on the theme of Measure ment of changing mechanical proper ties of carbon composite on nanosatel lite miniCube mission QB50 indepen dently under the guidance of the mas ter s thesis supervisor and using the technical literature and other sources of information which are all quoted in the thesis and detailed in the list of litera ture at the end of the thesis As the author of the master s the sis furthermore declare that as re gards the creation of this master s the sis have not infringed any copyright In particular have not unlawfully en croached on anyone s personal and or ownership rights and am fully aware of the consequences in the case of breaking Regulation 11 and the following of the Copyright Act No 121 2000 Sb and of the rights related to intellectual prop erty right and changes in some Acts In tellectual Property Act and formulated in later regulations inclusive of the pos sible consequences resulting from the provisions of Criminal Act No 40 2009 Sb Section 2 Head VI Part 4 In Prague July 30 2015 author s signature Contents List of Figures XIII List of Tables XV List of Codes XVII Lis
19. level One of the possible solutions how to increase an amplitude is to prolong the pulse to the coil In the Fig 8 8 are shown spectra of signals mentioned above At temperature 99 C and above the spectral peak at 970 Hz is suppressed and at 147 C the spectral peak at 300 Hz is suppressed as well co o o o B o Magnitude 20 log IY K Magnitude 20 log y Y K N o o 0 200 400 600 800 1000 0 200 400 600 800 1000 Frequency Hz Frequency Hz a Spectrum at 25 C b Spectrum at 52 C 70 s 60 o o o 60 o 99 2 2 o o 40 a a IM 3 30 3 2 E 20 20 E 10 0 i i 0 i 0 200 400 600 600 1000 0 200 400 600 800 1000 Frequency Hz Frequency Hz c Spectrum at 99 C d Spectrum at 147 C Fig 8 8 Spectrum comparison at different temperatures Frequency peaks were searched in narrow band of spectrum so the results in Fig 8 6 at higher temperatures are random Search ranges were basically established according to Tab 8 1 Tab 8 1 Searching ranges of peaks in the spectrum Line Range fl 120 Hz to 200 Hz f2 260 Hz to 330 Hz f3 700 Hz to 1000 Hz 8 MEASUREMENTS TESTING AND RESULTS o o 8 5 Final Implementation Final step of implementation was adding the operation system FreeRTOS to gether with CSP for further communication with OBC All previous progress of implementation
20. of this thesis 3 1 7 On Board Computer On Board Computer is synonymous for heart of probe Board is in charge of many functions but primary is operating other boards their requirements and reply to them Besides that it communicates with ground segment and organizing further actions on the deck o 3 1 Parts of Probe 3 1 8 Electronic Power System Power system consists of solar panels which generate electric energy and power board which transforms and stabilizes voltage at defined value 5 V or 3 3 V Board has two backup lithium batteries when probe is in solar shade This board is au tonomous and when power goes under critical value it cuts off all systems included OBC and then it waits for power m 3 1 9 Stabilising system It is necessary to stabilize probe to its defined state Small nanosatellites does not have rockets for stabilization but it has six coils two in each dimension Coils are excited by electric pulses and create the force necessary to stabilize the probe according to magnetic field of Earth 4 Mechanical Changes Measurement Main part of this thesis is about measuring mechanical changes Here on a few pages in this chapter is described theory about measurement of mechanical changes equations how to compute natural frequencies in the elementary case as a string or for the real model of a cantilever with one fixed side It is also described how to get the natural frequencies from
21. summer helped us not only with the project but with staying in Japan too Thanks to them we have progressed with our project and got great experience of Japan too Project VZLUSATI with cooperation VZLU consists of many measurements One is measuring of material aging in the space Measurement consists of exciting carbon fiber reinforced composite material by coil and measuring vibration by piezoelectric element Damped signal with exponential envelope is sampled by microcontroller and then are calculated resonance frequency and damping factor Using mentioned process is possibly to get final frequency resolution approx 0 12 Hz Every results as resonance frequency and damping factor as humidity and radiation are stored into memory and then are dispatch to the Earth during contact twice per a day Space probe CubeSat with name VZLUSATI will by launched in January 2016 and is funded by grants Technology Agency of the Czech Republic TA04011295 and TA03011329 ACKNOWLEDGEMENT Authors wish to thank Professors Minorikawa Ohsawa with the department of Mechanical Engineering and Professor Yasuda with the department of Electrical and Electronic Engineering Hosei University for their encouragement Authors also wish to thank Mr Saitoh and Iki with the department of applied informatics Hosei University for their technical support REFERENCES 1 Wide angle X ray imaging system with Timepix detector RIGAKU project number TACR TA
22. with high permeability It is able to conduct magnetic flux and pull itself closer to the coil minimize energy state Permalloy is the most suitable for this experiment if it is as small as possible to not affect the measurement and frequency spectrum of the cantilever It reckons with small deviation in frequency caused by permalloy 4 7 HM Panel When all these necessities are put together it leads to creating a sample of a panel with the cantilever The panel is called HM panel Final version is in the picture Fig 4 5 In the drawing position of the coil is shown and there is a permalloy under it Dimensions of the cantilever are 13mm x 67 mm x 1 mm Additional parameters of carbon fibre composite are Young s modulus which is E 34GPa and material density p 1700 kg m 4 8 Measured Parameters One of the most important parameters of the material is Young s modulus Additional parameter is an attenuation of oscillations during measurement 16 75 o 4 8 Measured Parameters o 26 A HM panel MCDP Axis MCDP Pant HM Coil Board N gt 2 Je ele i o o a 1 1 Coil core N lt o ls o V 2 81 36 9 dh n amp oU h O O 4 75 i PT1000 MCDP Pillar Stop Fig 4 5
23. 04011295 7 2014 12 2017 2 Experimental verification for space products and technologies on nanosatellite VZLUSATI project number TACR TA03011329 3 Cooperation and consultation with Czech Aerospace Research and Test Establishment 4 Ing L Sieger CSc consultations Czech Technical University in Prague Prague Czech Republic 5 K Yana Ph D consultations Hosei University Tokyo Japan 6 RNDr P H na CSc consultations Technical University of Liberec Liberec Czech Republic 7 Prof Ing P Sovka CSc Studying material for subject Digital Signal Processing Czech Technical University in Prague Prague Czech Republic 2013 2014 8 Decimation in time DIT Radix 2 FFT summer 2014 on line lt http cnx org contents ce67266a 1851 47e4 8bfc 82eb447212b4 7 gt 9 Datasheets for used parts CONTACTS M Urban urbanm24 fel cvut cz V Stehlikova stehlver fel cvut cz O Nentvich nentvond fel cvut cz D Content of DVD Folder structure of attached DVD Root Datasheets PCB Samples of signal Matlab files Nentvich Ondrej Diploma Thesis pdf
24. 14 but it is quicker and more suitable for calculation on Personal computer PC Microcontroller MCU etc How to get FFT from DFT is more described in the Martin Urban s thesis 1 Here is only one of the possibilities of implementing it to the Microcontroller 1 Nal _ 21rk Ye So ye IN 4 14 N r 0 Fast Fourier Transform has two phases one is bit reversing and second is the main computing process Bit reversing is shown in the Tab 4 2 where address of a point is swapped with the other This operation can be solved in the same memory space as original signal o 4 9 Computing Process of FFT 6 Tab 4 2 Bit reversing zo 000 000 o z 010 010 2 z3 011 110 6 za 100 001 2 Ts 101 101 5 ze 110 O11 3 a als ieee a FFT process is sums of points x as show in picture Fig 4 7 based on Cooley Tukey method Radix 2 and detail of one butterfly is in the following picture Fig 4 8 x 0 0 x 8 0 x 4 p 2 SL ZL V u BR W BR x 10 Ls x6 REA sip SSN x 1 WAN 8 W DAR co WV x 5 w XX fi DA xao PRA X 7 x 13 11 ws IAN x 3 ze AN 12 xy IND X 13 x 7 mo al 14 x 15 gt xas Fig 4 7 Flowchart of computing FFT with N 16 points A A A BW B B A BW W Fig 4 8 Detail of a butterfly 4 MECHANICAL CHAN
25. 36 37 XV List of Codes 6 1 6 2 7 1 1 2 7 3 7 4 7 5 Main loop of any task 36 Respond to Get Tempera ture Command 38 Parameters of Measurement 40 Decimation and bit reversing 42 Calculation of FFT 48 Calculation of attenuation 44 Output structure 46 XVII List of Acronyms ACK Acknowledge ADC Analog to Digital Converter CMOS Complementary Metal Oxide Semiconductor CSP CubeSat Space Protocol CubeSat Small standard nanosatellite of dimensions 10x10x10 cm per unit DFT Discrete Fourier Transform DiF Decimation in frequency DiT Decimation in time DK Data Keeper DSP Digital Signal Processing EEPROM Type of non volatiles memory EPS Electrical Power System etc etcetera FEM Finite element method FFT Fast Fourier Transform FIPEX Flux 9 Probe Experiment XIX FIR Finite impulse response Flash Type of non volatiles memory FRAM Type of non volatiles memory FreeRTOS Free Real Time Operating System FW Firmware HK House keeping HM Health Monitoring HW Hardware I O Input Output TC Inter Integrated Circuit INMS Ion Neutral Mass Spectrometer LCD Liquid crystal display LED Light emitting diode LHC Large Hadron Collider m NLP multi Needle Langmuir Probe MCU Microcontroller MOSFET Metal Oxide Semiconductor Field Effect Transistor NACK No Acknowledge OBC On Board Computer PC Personal computer ppm points per million Q
26. 90 os Zad 0059 23030 134108 030 8090 vas gt 8 03d vas v090 0 3 6 2 OS LOXL ad OSW ax 504 80 FOX Oud ZTVIX 4 vrszive SEEN 1 1 Schematic of power switch 49d OAT OWI HOS LOX 00XL0000 33 93d TOSIN LOXU 00X4 0090 DDA Sd ISOW 130 086 8100 0000 DDA 63 88 YIO 00007 04 00XL 8090 0090 Tod 0X 8090 2030 anov PO4 108 0MOX V000 8000 03d was w000 V000 som 10 vos 08d 32 800V ONS vox 55 9 03 oa ano AND 2 4232 02 10 g Fig A 1 YZOLADTEZ AE Er m 51565 Lyd 43 LOQV_ _ edad 0 38 03 020 12 6 559 HAV 59 Ivan os zsm JS OJJUODOJIDIN Agee 4819 10 94 OZ O I P180g i i i i i 1 y 99m so Py so i i i i i A Schematic of the board B HM board TOA 80 bisod MH eros so or MO Fig B 1 HM board B HM BOARD O 2 365 32 22 55 _ 24 4 lt 2 3 u LO o N O L Fig B 2 Dimensions of HM board C Article from Japan Measuring carbon fiber aging on orbit Bc Martin Urban Bc Veronika Stehlikova
27. B50 Missions of nanosatellites CubeSats RAM Type of volatiles memory SCL Serial clock SD Secure digital SDA Serial data SPI Serial Peripheral Interface SRAM Type of volatiles memory USART Universal synchronous and asynchronous serial receiver and transmitter vs versus VZLUsat 1 Marking of nanosatellite from institution VZL VZL Vyzkumny a zkusebn letecky stav a s Aerospace Research and Test Establishment List of Symbols A Fa Bn Cross sectional area m Elastic Modulus Pa Damping force N Inertial force N Reverse force N Quadratic torque section m Number points of signal Number of point least sguare method Own root of freguency equation Attenuation of system 57 Wavelenght m Angular frequency rad 5 Angular frequency of natural oscillations mas XXI Wn Co dec rak Js Angular frequencies of natural oscillations rad s Material Density ke m Cantilever width m Velocity of longitudinal waves ms Decimated frequency Hz Maximal frequency Hz Sampling frequency Hz Cantilever height m Quadratic sectional radius m Number points of moving averages Directive of line Spring constant Nm Cantilever length m Mass kg Li Yi Offset of line dB Velocity of propagation ms Point of signal mV Deviation m Time poin
28. CZECH TECHNICAL UNIVERSITY IN PRAGUE CESK VYSOK UCEN TECHNICK V PRAZE FACULTY OF ELECTRICAL ENGINEERING DEPARTMENT OF MICROELECTRONICS FAKULTA ELEKTROTECHNICK KATEDRA MIKROELEKTRONIKY MEASUREMENT OF CHANGING MECHANICAL PROPERTIES OF CARBON COMPOSITE ON NANOSATELLITE MINICUBE MISSION QB50 MASTER S THESIS DIPLOMOV PR CE 2015 Bc ONDREJ NENTVICH CZECH TECHNICAL UNIVERSITY IN PRAGUE CESK VYSOK UCEN TECHNICK V PRAZE FACULTY OF ELECTRICAL ENGINEERING DEPARTMENT OF MICROELECTRONICS FAKULTA ELEKTROTECHNICK KATEDRA MIKROELEKTRONIKY MEASUREMENT OF CHANGING MECHANICAL PROPERTIES OF CARBON COMPOSITE ON NANOSATELLITE MINICUBE MISSION QB50 M EN ZM N MECHANICK CH VLASTNOST UHLIKOVEHO KOMPOZITU NA NANOSATELITU MINICUBE MISE OB50 MASTER S THESIS DIPLOMOV PR CE AUTHOR Bc ONDREJ NENTVICH AUTOR PR CE SUPERVISOR Ing LADISLAV SIEGER CSc VEDOUC PR CE Prague 2015 esk vysok u en technick v Praze Fakulta elektrotechnick katedra mikroelektroniky ZAD N DIPLOMOV PR CE Student Bc NENTVICH Ond ej Studijn program Komunikace multim dia a elektronika Obor Elektronika N zev t matu M en zm n mechanick ch vlastnost uhl kov ho kompozitu na nanosatelitu miniCube mise GB50 Pokyny pro vypracov n 1 Prostudujte problematiku vyhodnocen tlumu exponenci ln tlumen ho sign lu vznikaj c ho kmit n m uhl kov ho kompozitu 2 Navrh
29. GES MEASUREMENT o The FFT process is divided into stages their number is log The flowchart Fig 4 7 has 4 stages of butterflies where on the left side is the bit reversed input signal and on the right side is the final spectrum Each butterfly is computed according to Fig 4 8 using equations 4 15 4 16 Complex points A and B represent each stage of butterflies and W eiT A A BW 4 15 By A BW 4 16 These eguations could be divided into real and imaginary part of numbers as shown in six following eguations 4 18 4 23 C BW R B jS B co dl sin are 4 17 y R C 5 7 sin R B cos 4 18 S C S B cos 3 R B sin 7 4 19 An R A R C 4 20 S An 55 4 55 4 21 R B 4 R C 4 22 S Bn 3 4 S C 4 23 One butterfly has six additions and four multiplication operations It is not so hard to calculate on small MCUs in case that values of sine and cosine functions are precalculated When all butterflies are computed the last step is to get values of every frequency of the signal This is performed in absolute value as in the equation 4 24 12 14 z YR 2 5 2 4 24 4 10 Computing Process of Attenuation Envelope of the signal represents attenuation By using signal processing damp ing constant Attenuation of system is extrapolated 20 75 4 10 Computing Process of Attenuation 6 The process co
30. S E 0 Ho E E gt gt gt 200 3 i 400 0 0 5 1 1 5 0 0 5 1 1 5 0 0 5 1 1 5 1 5 1 5 1 5 a Sampled signal with too b Sampled signal with ideal Sampled signal with too short pulse width pulse width long pulse width Fig 8 5 Differences between pulse widths 8 4 Different Climate Conditions The creation of a reference for further processing is necessary to measure all boards and parts of the experiment in different climate conditions The HM board and panel were tested only in temperature oven not in the vacuum because the chamber is not big enough for HM panel with frame hanged on a wire 5 1000 f2 f3 attenuation 900 800 700 30 600 7 Frequency Hz Attenuation 400 4 5 0 100 1 10 20 40 60 80 100 120 140 160 180 Temperature C Fig 8 6 Chart of different climate conditions 8 MEASUREMENTS TESTING AND RESULTS o o We made climate testing only in heated oven This test was performed in tem perature range from 25 C to 176 C as shown in the chart Fig 8 6 where frequency of first natural frequency fl line is almost unchanged in the whole range of measure ment conditions Second line f2 is an now unspecified frequency peak at 300 Hz probably is natural frequency of the frame Its changes began at temperature around 140 C Third frequency f3 should not be here it is mirrored from 1050 Hz Th
31. aded and other variables are read from the boards measurement process can start Microcontroller has one dedicated timer for timing of excitation coil which stops excitation Also one timer is dedicated for sampling of signal with higher priority of interrupt for precision timing All points are sampled and averaged four times and then stored into SRAM to the space dedicated for sampled signal Number of points is 4096 at fs 4kHz It covers a little bit more then 1s of signal From this determination of Sampling frequency is possible to have spectrum up to 2 kHz 7 4 Signal Processing FFT The most important part of signal processing for our purpose is computing FFT to find out resonant frequencies of sampled signal More detailed description about meaning of looking for frequencies is in chapter 4 8 1 Introduction to the FFT process is in Martin Urban s thesis 1 and in chapter 4 9 with all equations for calculations Here I will only simply describe how to calculate it in 8 bits MCU and the implementation into the Microcontroller The whole FFT computing process has two parts address bit reversing and the actual FFT process m 7 4 1 Decimation and Address Bit Reversing Decimation is used for reducing the spectrum of signal but Shanon Niquist theorem of fs gt 2fmax for searching frequency must be followed in this case fmax With the same length of window for computing FFT resulting spectrum has better resolution
32. and HM panel 0 OK 1 problem uint8_t xtal 1 ztal works uint8_t sram di sram OK uint8_t loaded_MB 1 loaded temperatures from Measure Board uint8_t loaded OBC time 1 loaded time from OBC uint8_t config out dec 1 configuration is invalid decimation 0 or gt 4000 Hz applicated 1000 Hz uint8 t config out freg 1 configuration is invalid any of the frequencies is out of range greater then decimate_to_f Hz 2 applicated decimate_to_f 2 Hz uint8_t config_out_ls 1 configuration is invalid least square averaging is out of range 200 1000 points applicated 200 Hz uint8_t config_out_power 1 time of power switch is zero default value is 10 uint32_t OBCTime time of measurement int16 t f1 3x freguencies int16 t f2 int16 t 3 int16 t attenuation directive k directive of attenuation from equation y kx q int16_t tempADT temperature from ADT thermometer int16 t templ 6x temperature from HM panel int16 t temp2 int16 t temp3 int16 t temp4 int16 t temp5 int16 t temp6 int16 t orientation1 5x orientation of probe int16 t orientation2 int16 t orientation3 int16 t orientation4 int16 t orientation5 hm output data t 46 75 8 Measurements Testing and Results It is necessary to verify the final device in many climate conditions such as temperature changes pressure etc same as repr
33. assembly of any board it is required to verify functions of the each measurement system Verification of HM board and panel was performed in stable environmental conditions in VZL in room with stable temperature around 23 5 C and about 80 of relative humidity with small oscillations of these values Many measurements in a row were performed in these conditions Resulting spectra were similar to Fig 8 1a and results of first natural frequencies were the same 173 3 Hz in all cases The second natural frequency oscillates between 1 060 5 Hz and 1063 Hz Median of the frequency range is at 1 062 Hz 54 75 o 8 7 Conclusion Oscillations around this value are probably caused by the quality of the signal or noise in the spectrum which can be seen in the Fig 8 18 Histograms of these natural frequencies are in pictures in Fig 8 9 Counts Counts 9 173 3 1060 5 1061 1061 5 1062 1062 5 1063 Frequency Hz Frequency Hz a Histogram around 173 Hz b Histogram around 1062 Hz Fig 8 9 Histograms of natural frequencies Second examined parameter is the attenuation of signal Their results were quite unexpected during a measurement in the same conditions The attenuation is decreasing in absolute value according to Fig 8 10 and it does not depend on the temperature At the beginning of measurement the temperature goes down and than up but attenuation still goes down in its absolute value Maybe in ca
34. ation storing and bitreversing raw signal into spi memory for u int16 i 0 i lt NO_POINTS i position pgm read dword far kbitrev i decimation position long position DECIMATION FACTOR if decimation position lt MEM SIGNAL POINTS read data spi mem read word decimation position 2 MEM SIGNAL BEGIN signal offset point real read data else point real 0 store into SPI memory spi_mem_write_complex long MEM_FFT_BEGIN i 8 point Where DECIMATION FACTOR is ratio of fs and face MEM SIGNAL BEGIN MEM_ FFT BEGIN are positions of beginning address of SPI memory NO_POINTS is num ber of points of sampled signal bitrev is an array of bit reversed addresses and signal offset is value for elimination of DC part of signal 12 42 75 o 7 4 Signal Processing Attenuation 6 B 7 4 2 FFT Process FFT process is based on Cooley Tukey method of computing according to egua tions 4 18 4 23 Functions sine and cosine are precalculated in a table instead of computing it in MCU to save time Final spectrum is represented as an abso lute value of complex point 4 24 For our purpose it is necessary to know only the maximum peak and from this point of view is sguare root omitted to save time of cal culations The code used for computing the FFT is in following listing Code 7 3 13 Code 7 3 Galculation of FFT msi s actual state angf NO_POINTS 2 twiddle fac
35. d to know the characteristics of the whole satellite and how to affect the resulting spectrum of sensed signal If the construction of the probe produces another natural frequencies the piezo should show it in the case of strong signal The MCU must be prepared for each situation to solve it by setting parameters of measurement These tests will be performed in September 2015 unfortunately after finishing this thesis Also the probe will be tested in the whole range of operation temperatures and climate conditions 9 Conclusion The small satellites from CubeSat family are useful for University teams to build them for their low price Probes have a quite great portfolio of basic products such as OBC power systems solar panels chassis etc Teams can use these products or develop their own but must observe some specification like dimensions weight etc From this point of view a new technology for further missions can be developed During this mission sensors materials and devices will be tested in space condi tions For example testing carbon fibre material for radiation shielding mechanical quality and evaporation from the material Another example is verifying the Lobster Eye optics together with Medipix Timepix detector which detects X ray radiation from the stars but mainly from the Sun This master s thesis Measurement of changing mechanical properties of carbon composite on nanosatellite miniCube mission QB50 describes one o
36. different types of sensors in the probe For these sensors was necessary to design the PCB Printed Circuit Board too Fig 3 Printed circuit board for the measurement III INPUT SIGNAL Input signal is measured by piezoelectric element glued on composite plate and has approximately exponential envelope modulated by material resonance frequency Exponential envelope is caused by attenuation in the material Useful signal length depends on attenuation which is caused by physical dimensions and material properties in the material and amount of energy excited by coil In the picture fig 4 is shown one example of input signal Measured real signal has total length 0 9 s C ARTICLE FROM JAPAN Input signal is sampled by 8 kHz In case of computing attenuation envelope is used all signal for the better resolution in time This signal is averaged by moving averages Smoothed result is logarithmed and attenuation is directive of the course o a 5 a a tn 8 Relative Amplitude 1 1 1 1 L L a a i A 1 o 01 02 03 04 05 06 07 08 09 Time s Fig 4 Measured real damped signal Resonance frequency is computed from decimated original signal using Fast Fourier Transform If we use only FFT on the original signal we can calculate only with 1024 points length of window because we are limited by memory size installed in the probe Computing by this process has final resolution more than several Hertz
37. ega128A4U from Atmel company Microcontroller is responsible for signal sampling from piezo and excitation of coil assembled on HM panel HM_board_1 5_TOP ON 10 02 2015 Power JE switch Piezo connection Thermometer MCU SRAM Fig 5 3 Picture of HM board with highlighted parts m 5 2 1 Microcontroller for Payloads Microcontroller ATxMega128A4U has been chosen for payloads which is one of many compatible devices with CSP It is guite powerful MCU for simple tasks such as signal sampling temperature measurement communication with other boards etc Basically it is focused on easy tasks but for measuring mechanical changes it is necessary to get the freguency One of the possibilities how to find out the freguencies is to calculate period Using this method is not too much reliable for more then one freguency contained in sampled signal So it have to calculate FFT It is not too much suitable for this 27 75 5 DEVICE FOR MECHANICAL CHANGES MEASUREMENT M o o purpose because of the computing power but with more time for calculating it is possible to do it The best way to process FFT is use a DSP Microcontroller It will calculate it in shorter time because it has specialized instructions for it The MCU ATzMega128A4U is from AVR family of chips and value 128 signifies the size of internal Flash memory it has 128 kB of it Some selected parameters of the Microcontroller are in the table Tab 5 1
38. ent on nanosatel lite M miniCube mission QB50 master s thesis Prague Czech Technical University in Prague Faculty of Electrical Engineering Department of Microelectronics 2015 35 p Supervised by Ing Ladislav Sieger CSc 3 CubeSat online 2015 cited 2015 04 14 Available from https www qb50 eu 4 GomSpace online cited 2015 04 14 Available from http gomspace com 5 Medipix online cited 2015 07 16 Available from http medipix web cern ch MEDIPIX 6 VZLUSAT 1 Czech 2U Cubesat for IOD s online cited 2015 04 15 Available from http www itu int en ITU R space workshops 2015 prague small sat Presentations VZLUSAT1_ITU pdf 7 Manual for the FIPEXonQB50 Science Unit PART Flight Model online 2014 12 01 cited 2015 06 16 Available from https www qb50 eu index php tech docs category 19 up to date docs download 84 fipex user manual 8 PODU KA J Experiment ln anal za dynamick ho chov n vetknut ho nos n ku Brno Vysok u en technick v Brn Fakulta strojn ho in en rstv 2011 36 s Vedouc bakal sk pr ce Ing Luk B ezina Ph D 9 BREPTA R P ST L a TUREK F Mechanick kmit n Vyd 1 Praha Sobot les 1994 589 s esk matice technick Sobot les ISBN 80 901684 8 5 10 HANA P INNEMAN A D NIEL V et al Mechanical properties of Carbon Fiber 3 Composites for applications in space Proc SPIE 9442
39. etween command and return value is approx 250 ms This time is derived from the length of conversion which is about 240 ms All parameters are from datasheet 18 6 Communication with Other Boards Communication is one of the most important things Without it data cannot be transmitted and boards do not know what to do In this case board communi cates with the rest of the probe connected to the same bus via I C interface All boards have specific communication protocol called CubeSat Space Protocol It is an universal protocol specially for CubeSats 6 1 Interface Generally I C interface is connected via two wires SDA and SCL with common ground The bus required two pull up resistors usually in range of 470 10k0 It depends on specification from the manufacturer With more devices on the bus it is recommended to reduce the resistance When bus is idle positive voltages are on SDA and SCL wires Visualization of the connection is on the follow figure Fig 6 1 Resistors Rpy are not necessary at every device one on each wire is enough ADC DAC HC Master Slave Slave Slave Fig 6 1 Schematic of connection I C devices 19 Communication protocol between master and slave unit is shown on the next figure Fig 6 2 Communications starts with start condition SDA falling edge prevents SCL falling edge Then follows 7 bits address and last bit is read high 33 75 6 COMMUNICATION WITH OTHER BOARDS o o
40. external oscillator with requested frequency of the signal with defined output amplitude 26 Computing process of attenuation is in the chapter 4 10 Resulting attenuation was sent to the computer and then saved into a file After that evaluated and compared with theoretical value The difference between the value computed by MCU and Matlab was minimal better than 0 5 compared with single and double precision of calculations For example Matlab computed attenuation as 4 08 vs MCU as 4 07 8 3 Testing of Pulse Width Examination of right pulse width is necessary to determine a proper waveform which is not clipped and must have maximal amplitude of 400mV as mentioned before in chapter 5 2 5 Waveforms in Fig 8 5 shows the comparison of differences of signal in dependence on pulse width The Fig 8 58 was sensed with 0 5 ms pulse width Fig 8 5b with 1ms and Fig 8 5c with 2ms From this point of view it is best to excite the coil by 1 ms pulse width In the last picture is shown how an internal diode of MCU clipped the sampled signal in negative values Positives values has normal amplitude This disadvantage could be solved with additional voltage reference at half of ADC range in our case the reference should be 3 3 1 6 2 1 03 V Supply voltage is 3 3 V and the division factor 1 6 is chosen as an internal reference of ADC 15 50 75 o 8 4 Different Climate Conditions 400 200 s
41. f orbit where the probe will be The height of the orbit will reduce over time and data speed will be slower Due to limited volume of data which is possible to send is necessary to process raw data on orbit and let to the Earth arrive only results Carrier with our VZLUSATI will start January 2016 from Brazil with an Ukraine rocket It will be brought to low orbit 350 km high Then it will collect data at least two month It depends on how lucky will the probes be during the launch in case of a great starting angle it could work half year too Of course there are many other possibilities which can involve lifetime of satellite It is not sure how well will it deal with temperature changes on orbit with radiation exposure and other unexpected conditions Whole time it will send data to the Earth and in the end space probe will burn in atmosphere INTERNSHIP AT HOSEI I SUMMARY Our work comprises technical solution of measuring aging of carbon fiber reinforced composite and as an usual projects at CTU it is quite complex It begins from theory mathematical and physics modeling and working out the tasks in simulation programs Then there is a part of trying to apply proposed resolutions on testing facilities only to check the possibility in real situation Simulations include these with high sophisticated software still cannot cover the entire range of possibilities that may affect the device as a whole If the device on a b
42. f the many ways how to measure Young s modulus of elasticity for verification of quality of any mate rial In this case natural frequencies were gained using FFT and from them modulus of elasticity was retroactively calculated Additional parameter is a damping factor attenuation It serves to better spec ify a model of HM panel using FEM and for computing of all parameters retroactively and with a greater accuracy The whole measurement process in HM board is controlled by the main Micro controller which has to communicate with other boards mainly with OBC This connection must be perfect and reliable because when it fails the measurement cannot be started anymore Same requirements apply for assembled parts on the boards They also must be robust enough to resist radiation from space and in some cases low or high temperatures For mission on low orbit these conditions are not so strict but they should be observed for further missions where worse conditions then on low orbit may occur Tests of reliability of communication with OBC were performed on many com mands which were sent to the MCU on the HM board Responses were positive almost every time but once had happened that HM board stopped to communicate and had to be rebooted This was performed by switching power supply off and then on again After this step the MCU communicated normally This could happen during an unexpected situation on the bus by coincidence 57 75 9
43. frame 47 75 8 MEASUREMENTS TESTING AND RESULTS o o because the left chart does not have them Health Monitoring panel mounted in jaws at the fixed side of cantilever and it may be considered a reference sample for further evaluation Spectrum above Sampling frequency 2 does not affect main part of it only smooths it a little This statement is based on using Finite impulse response FIR filter on sampled signal where differences in spectrum with and without using FIR filter are minimal Comparison of them is in the pictures Fig 8 2 100 100 lt lt 80 80 2 2 ES o 60 60 N N 3 40 5 40 S 20 S 20 0 i 1 0 0 500 1000 1500 2000 0 500 1000 1500 2000 Frequency Hz Frequency Hz a Spectrum of signal without using FIR Spectrum of signal with using FIR filter filter Fig 8 2 Differences in spectra without and with using FIR filter 8 1 Beginning of Research The main part of research has began in the spring of 2014 when the highest possible resolution in frequency was expected around 0 1 Hz At the beginning of research we took only the part to 250 Hz where the main peak the first natural frequency of the cantilever is located All these claims are based on similar spectrum to reference Fig 8 18 which was sensed by oscilloscope for the first time At first it was required to learn about the used Microcontroller ATrmega128A4u so we got develo
44. g In this case the following equation 4 8 for examination Young s modulus of elasticity is applied E v 4 8 where Velocity of propagation v is defined as v Af A is Wavelenght F is Elastic Modulus p is Material Density In this case it is possible to determine that A 4L where L is part of string length illustrated in Fig 4 3 U A Fig 4 3 Length of string L vs wavelength A 4 3 Cantilever Oscillations For our purpose to approximate to the real situation which is in the simplest case cantilever with one fixed side and one freely hanged as is in the picture Fig 4 4 Natural frequencies of cantilever are described by a few equations In the first step it is needed to find own roots of frequency equation of the cantilever 4 9 cosh Bnl cos Bn 1 0 4 9 13 75 4 MECHANICAL CHANGES MEASUREMENT o Y Fig 4 4 Picture of the Cantilever This equation can be solved only by numeric method and its four first roots are Bl 1 875 Bol 4 694 Bal 7 855 Bal 10 996 where Bn is Own root of frequency equation l is Cantilever length 8 Natural frequencies are computed using following equation 4 10 For this for mula it is required to know computed roots and some other parameters all necessities are in Tab 4 1 and put into equation 4 11 2 6 2 Coj 4 10 JE IJ N ry 4 11 BD E Jh gt 21 412 where j is Quadrat
45. g new on the orbit without risking lots of funds These small satellites are carried on next to a main standard satellite a little bit like a stowaway which is the carrier primary used for So these nanosatellites like SpaceCube are depending on projects of standard space research without the possibility of an independent start As a small satellite it does not have own active power also Fig 1 Appearance of the cube sat VZLUSATI C ARTICLE FROM JAPAN In general the main part of people working on SpaceCubes recruits among the college students and their mentors It is a chance to work on an interesting project with a possibility of future extension Itis our case too After the launch we will cooperate with University of west Bohemia in Pilsen where will be the headquarters for communication with orbit We will get the raw measured data from them that we will process into final results II VZLUSATI IN GENERAL Project VZLUSATI carries ten experiments in total These experiments have various aims There are two main tasks on the satellite One tests a new carbon fiber reinforced composite panel with metallic coating made by 5M and TTS companies for utilization on space probes from the perspective of strength weight radiation shielding durability and evaporation the second one is Wide angle X ray imaging system Examining of this composite panel could lead to verification of suitability for this kind of material
46. h ports and its address is 6 Between basics functions of all board are ping that board is still alive and House keeping HK data it returns information about board Between user defined commands are for example return data signal etc All functions are in the table below Tab 6 2 6 4 Communication with HM Board 9 Tab 6 2 Communication ports of HM board Port number Description 18 Measure starts 19 Returns signal to DK 20 Returns results of measurement 21 Returns temperature of HM board Meaning of commands from table Tab 6 2 are as follows 6 4 1 Measure Starts Measuring starts only when HM board gets packet to the port number 18 Packet could be empty or contain anything but it does not matter to the program what came Response to the originator is simple ackMes as text When signal is sampled all that is calculated returns results to the DK m 6 4 2 Returns Signal to DK When this command is called MCU returns signal of previously sampled to the DK divided into many chunks each with predefined size 32 Bytes When signal is transmitted to the ground there it will be reconstructed E 6 4 3 Returns Results of Measurement This command returns only results that have been calculated previously It means the same thing as at the end of command Measure starts This command is redundant but it serves as a backup when something goes wrong during performing command Measure starts m 6 4 4 Return
47. he for cycle does not calculate with time because it would mean additional time for calculation Recalculation to the time is performed after the for loop using reciprocal value of fs Final results of least square method are in reg_k which is a directive of line our attenuation and in reg_q is an offset of line Only directive is stored offset is not stored because it does not corresponds to the maximal value of original signal and also it is not necessary for further evaluation 7 6 Store Results All measured and received data will be saved to the DK with the following structure Code 7 5 and than transmitted to the earth for next process 7 7 Conclusion The small MCU ATrMEGA128A4U is able to compute FFT with 8192 points but it takes a lot of time approx 15s because the code is not very optimized for speed On orbit there will be a lot of time for this task it is established that the shortest delay between measurements will be at least 5min So 15s is enough with a huge reserve it is not calculated with other delays from getting the temperature time from OBC and orientation of the probe Totally it is estimated on around 20 30s from start command to saving the results When is required also raw signal to this time is added approx next 15s to store the data 7 MEASUREMENT PROCESS IN HM BOARD o o 5 o o o 2 Code 7 5 Output structure typedef struct __attribute__ packed status of HM board
48. hen the amplitude is reduced to the appropriate value For further using of differential input of ADC is better to shift up virtual zero negative input to the half of reference voltage to approx 1V 30 75 o 5 2 HM board E 5 2 6 Thermometer Health Monitoring board is equipped with one thermometer ADT7420 commu nicating through I C interface Its resolution is 16 bits with sign In the following table Tab 5 2 are shown other basic features Maximum allowed number of these devices is four because this chip provides four different addresses Address can be changed by two pins 40 A1 on the package Rest is set by the manufacturer and the address range in hexadecimal format is from 0x48 to 0x4B Pins 40 47 on HM board are connected to the ground logical zero thermometer ADT7420 has address 0x48 Tab 5 2 Parameters of thermometer ADT7420 18 Parameter Value Description Resolution up to 16 bits Maximal resolution of thermometer Temperature range 20 to 105 C Temperature range of thermometer Precision 0 25 C Accuracy of thermometer in temperature range from 20 to 105 C Supply 211009 Power supply voltage range Thermometer is connected through C interface and the bus required external pull up resistors with resistance about 4 7 kQ Value from the thermometer can be get by one CSP packet if any board needs to know it Parameters of the communication are specified in chapter 6 4 4 Delay b
49. ic sectional radius co is Velocity of longitudinal waves J is Auadratic torgue section and A is Cross sectional area Conversion between angular speed and freguency is in following eguation 4 13 La gt 4 13 14 75 4 5 Measuring by Piezo Tab 4 1 Cantilever Parameters Symbol Description l Cantilever length b Cantilever width h Cantilever height E Elastic Modulus p Material Density Investigated parameter is Young s modulus of elasticity F and it is verified using formula 4 11 From this point of view it is a prereguisite that the other parameters do not change according to Tab 4 1 Properties of cantilever could be also simulated using FEM This issue was sim ulated by Petr H na from Technical University of Liberec using FEM and results of this method were almost the same as using equation 4 11 8 10 For verifying natural freguencies it is needed to get oscillations from the beam Signal measurement could be performed by two methods One is by accelerometer and another one is by piezoelectric plate 4 4 Measuring by Accelerometer Accelerometers are usually small and lightweight about a few grams One disadvantage is that they must be placed into position with the highest variation of signal for the highest acceleration That location is at the end of the cantilever It is not recommended for this purpose because the weight of the plate is about a few grams and accelerometer wi
50. ighest mechanical changes Assembled HM panel is in the picture Fig 5 1 Surface of the panel is covered by two reflective materials There is nickel com pound on the left side and golden compound on the right side These materials should reflect thermal radiation for example from the Sun Health Monitoring panel has six sensors PT1000 which measure temperature transmission through the panel Temperature is also important factor for calibration of measurement because properties of carbon fibre material could change Final assembling of these sensors are shown in the pictures Fig 5 2 a Front side b Back side Fig 5 2 Detail of HM panel All temperature sensors are connected to the measure board which measure tem perature from almost all PT1000 Board could be called by any payload through C interface via CSP Wires from coil and piezo are connected to the HM board which processes all measurements of mechanical changes Piezo is connected to the connector marked as Piezo and coil is connected to the connector called Coil o 5 2 HM board 5 2 HM board Health Monitoring board is shown in the picture Fig 5 3 which consists of parts such as main connector which is not standardized for CubeSat because of the weight and also board is the last one in row in the probe So it is not necessary to connect all signals to the board only power supply and I C interface Next part is external oscillator connected to the MCU ATrM
51. in frequency Sampled signal has default fs 4 kHz and decimation factor is 4 It is according to new sampling frequency 1 kHz Decimation factor could be changed by the setting stored in DK decimation factor But if the factor is one decimation will not be used Bit reversing is one of the requirements for computing FFT when is used Dec imation in time DiT or Decimation in frequency DiF method Main difference between DiF and DiT is in order of performing bit reversing and FFT process Dec imation in time consists of this bit reversing and then FFT process and Decimation 41 75 7 MEASUREMENT PROCESS IN HM BOARD o o in frequency has reversed order Process of bit reversing is simple Take one point of signal with known address swap all address bits and store the point into this changed address In the table Tab 4 2 is visualised this claim addresses for example are in binary form in length of 3 bits From this point of view it is recommended to use DiT method and combine advantageous decimation with bit reversing Every point is loaded and stored one time from into the memory For simplification of reversing all bit reversed addresses are stored in array in Flash memory of MCU In the following code listing Code 7 2 is implementation of decimation and bit reversing Code 7 2 Decimation and bit reversing void decimate and store void here is variables and verification of decimation frequency in configur
52. is fact is caused by undersampled signal only 2 kHz sample rate on testing equipment This test was performed at time when we were determining sampling frequency and method how to get the most precise frequency 0 1 Hz resolution of the first natural frequency The frequency line f3 has began to change significantly at 100 C This could be affected by the quality of glue hardening at this temperature glass transition temperature of glue has been reached The attenuation has been affected already at lower temperature of 75 C then the attenuation is changed Also signal level at temperatures over 100 C was quite low so it affected frequencies as well as the attenuation 400 U mV U mV 400 i i 02 04 06 08 1 0 02 04 06 08 1 t s t s a Signal level at 25 C b Signal level at 52 C 400 r r r r 400 200 200 400 1 i 1 i 400 1 i 1 i 0 0 2 0 4 0 6 0 8 1 0 0 2 0 4 0 6 0 8 1 t s t s c Signal level at 99 C d Signal level at 147 C Fig 8 7 Signal level comparison at different temperatures 52 75 o a 8 4 Different Climate Conditions o In Fig 8 7 are shown charts with different temperatures It is guite certain that at higher temperatures the signal is weak The nanosatellite should operate probably up to 80 C where signal is strong enough Results over 90 C are useless because of signal
53. ith N 16 points Detail of a butterfly Result of moving averages Attenuation with directive Picture of HM panel Detail of HM panel Picture of HM board with highlighted parts Schematic of power switch Schematic of piezo connection Schematic of connection IC devices 19 Chart with communication via I C 19 CSP header Difference in spectra with different placement Differences in spectra without and with using FIR filter Photo of development kit 24 Signal with triangular envelope 5 11 12 13 14 17 18 19 19 21 22 25 26 27 29 30 33 34 34 47 48 49 XIII 8 5 8 6 8 7 8 8 8 9 8 10 A1 B 1 B 2 Differences between pulse widths Chart of different climate conditions Signal level comparison at different temperatures Spectrum comparison at different temperatures Histograms of natural frequencies Chart of temperature and attenuation in time Schematic of power switch HM board Dimensions of HM board 51 52 53 59 59 65 67 68 List of Tables 3 1 Scientific units 3 4 1 Cantilever Parameters 4 2 Bit reversing 5 1 Parameters of MCU ATxMegal28A4U 15 5 2 Parameters of thermometer ADV 420 18 secos 6 1 xTaskCreate Command Parameters 22 6 2 Communication ports of HM board 8 1 Searching ranges of peaks in the spectrum 15 19 31
54. iversity in Prague CTU In this project also cooperates many other companies for example Rigaku Innovative Technologies Europe s r o 5M s r o TTS s r o Innovative Sensor Technology s r o DENTEC and with many other individuals and specialists The goal of it is to carry on the orbit a nanosatellite which will perform diverse experiments The 3850 project relies on building satellites of defined measures based on cubes 10x10x10 cm which can be combined up to three cubes in a row for one probe this time In the future should be constructed bigger satellites up to 2x2x3 units Their conformity allows to carry a large number of different probes to the orbit together The reason of the project s name is that there will be space for 50 units of two cubes 2U standardized units aboard One of each 2U have to had maximum power supply up to 2 Watts and up to 2 kg The philosophy of this project is to make a path for cheaper and easier manufacturing of satellites in the future To build and carry a satellite to the orbit is not a cheap thing so it is the reason why even on high end projects are still used old verified materials and technologies They simply are proved by time and the project sponsors do not want to risk using any new and untested parts due to which the whole project could crash Compared to these large separate projects the SpaceCube program offers an opportunity for lots of scientist who need to test somethin
55. lored lower thermosphere measuring or research in these altitudes about 200 380 km Launch is scheduled on February 1 2016 from Alcantara launch site in Brazil by rocket Cyclone 4 3 2 1 One CubeSat unit One unit of the probe has these parameters 3 e Dimmensions 10 x 10 x 10cm per unit e Weight up to 1 kg per unit e Units up to 3 units in row 30 x 10 x 10cm during this mission Advantages of CubeSats are that existing standardised HW boards like On Board Computer OBC Electrical Power System EPS board solar panels radio board etc are qualified for these missions and use in space environment It is only neces sary to implement communication with other boards 4 3 CubeSat VZLUsat 1 X ray optic Deployable Solar panel HM panel E HKR board HMboard gt HRRA Measuring mN board XRB diodes Antenna Volatiles X ray detector Radio a Picture of deployed satellite b Measurements on satellite Fig 3 1 Nanosatellite VZLUsat 1 One of the Czech nanosatellites is VZLUsat 1 shown in Fig 3 1a Vyzkumny a zku ebn leteck stav a s Aerospace Research and Test Establishment VZLU is supervisor and main coordinator of project and it has whole team participating on the probe The team involves some universities and companies such as
56. ment is 18 as written in the Tab 6 2 Process of the main measurement is divided into several steps Read parameters of measurement from DK Get temperatures from all sensors from Measure Board Get orientation of the probe Get actual board time Excitation of coil for a defined time loaded from DK Sampling of signal Signal processing FFT Signal processing attenuation Store data to the DK O 6 U A 2 7 1 Read Parameters of Measurement Parameters of measurement are stored in DK storage number 1 Configuration contains mainly parameters around computing Young s Modulus of Elasticity such as range of frequencies where natural frequency is looked for decimation frequency pulse length to the coil and number of points of least square method from the beginning of signal Whole structure of configuration is in listing Code 7 1 7 MEASUREMENT PROCESS IN HM BOARD o o Code 7 1 Parameters of Measurement packed 3x tolerance range of frequencies uint16_t f1_min uint16_t f1_max uint16_t f2_min uint16_t f2_max uint16_t f3_min uint16_t f3_max typedef struct __attribute_ how long is power switch on current to the coil uint8_t time_power_switch_on to which frequency is decimated to uint16_t decimation_factor length of least squares uint16 t length of ls Jhm config t Accuracy of frequency first six variables is in Hert
57. n te algoritmus vyhodnocen tlumu m en ho sign lu 3 Prostudujte vhodn zp soby excitace uhl kov ho kompozitu a sn m n jeho kmit 4 Navrhn te syst m pro m en zm n mechanick ch vlastnost uhl kov ho kompozitu v z vislosti na zm n teploty 5 Relizujte vhodn zapojen z 3 a 4 pro Payload nanosatelitu miniCube mise QB50 6 Realizujte m en pod syst mem RTOS 7 Ov te a zhodno te funk nost syst mu Seznam odborn literatury 1 JAN J slicov filtrace anal za a restaurace sign l 2nd ed Brno VUTIUM 2002 427 p ISBN 80 214 2911 9 3 T MA J Zpracov n sign l z skan ch z mechanick ch syst m u it m FFT Praha Sd lovac technika 2000 168 p ISBN 80 901936 1 7 3 HANA P INNEMAN A DANIEL V et al Mechanical properties of Carbon Fiber 3 Composites for applications in space Proc SPIE 9442 Optics and Measurement Conference 2014 2015 no 1 DOI 10 1117 12 2175925 Vedouc Ing Ladislav Sieger CSc Platnost zad n 31 6 2016 L S prof Ing Miroslav Hus k CSc prof Ing Pavel Ripka CSc vedouc katedry d kan V Praze dne 16 2 2015 Czech Technical University in Prague Faculty of Electrical Engineering Department of Microelectronics Master s Thesis Assignment Student Bc NENTVICH Ond ej Study program Communication Multimedia and Electronics Focused Electronics Topic Measurement of changing mechanical properties of carb
58. nected directly to the OBC Everything about humidity sensors and measuring evaporation is described in detail in thesis of my colleague Martin Urban in 1 3 1 5 Radiation Shielding Measurement The goal of this experiment is to verify quality of radiation shielding of Carbon fibre composite using three same XRB diodes One is looking into the space second is covered by this material and third is shielded by composite and by tungsten sheet Evaluation of shielding quality is comparing all three diodes together First serves reference for maximal intensity of radiation middle diode real intensity of radiation and third is measuring background This issue is described in more detail in thesis of my colleague Veronika Stehl kov in 2 m 3 1 6 HM System Health Monitoring system contains two parts One is mechanical part with beam made of carbon fibre composite and six thermometers PT1000 Thermometers measure heat transmission from one surface to another and how good reflectivity of thermal radiation surface layers of Nickel or Gold is Also HM panel measures aging of carbon fibre material using excitation of beam and measure its oscillations by piezo Whole signal is sampled by ADC in MCU Microcontroller is on the HM board which is the second part of Health Monitoring system Evaluation of signal is done by FFT and results are resonant frequencies of beam and of the probe This issue is described in more detail on followings pages
59. nsists of a few steps 1 Rectify signal 2 Compute moving averages 3 Logarithm signal 4 Compute least square method 4 10 1 Rectify and Moving Averages Signal rectification is quite simple task It is only makes an absolute value over the sampled signal When it is performed the next step is to compute moving averages which is only sum of points in time divided by number of point Expression for moving averages is in equation 4 25 1 k i r ieZ 1 0 N k 4 25 Where z is Point of signal N is Number points of signal k is Number points of moving averages The result of this step is in the following chart Fig 4 9 Signal length is N 4096 points and number of moving averages is k 512 0 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 1 5 Fig 4 9 Result of moving averages 4 MECHANICAL CHANGES MEASUREMENT o W 4 10 2 Logarithm Moving Averages Attenuation of damped oscillation is in ideal state exponential We will suppose that for real measured signal Expression of any damped oscillated signal could be this 4 26 with one modulated frequency Simulated signal by Matlab in Fig 4 2 has similar waveform to real measured signal Fig 4 6 but real signal has more then one frequency and this is only parable to this situation y Asin wt e 4 26 Using moving averages to the simulated damped signal gets something similar to the Fig 4 9 When natural logarithm is used
60. obe VZLUsat 1 has X Ray optics with focal length approx 20cm and should be looking into the Sun This state will happened twice a year because of static orientation of probe For detection of the right orientation there are three sensors two in UV spectrum and one in IR spectrum Main reason why probe has two UV diodes is that one is sensitive in maximum radius approx 80 second one has radius reduced to approx 15 by a small tube The first one is looking for sources of UV radiation mainly from the Sun in wide angle and the second one has narrower angle for more precise determination of Sun position When both sensors have strong enough UV signal then Medipix will start up which then take a picture of the Sun Medipix is CMOS silicon detector originally developed as low energy 1 20 keV X Ray detector for Large Hadron Collider LHC in CERN Board with this chip assembled on the probe was used as medical equipment for scanning soft tissues with high resolution For this mission Medipix Timepix is used for detection X Rays from the Sun 5 o 3 1 Parts of Probe E 3 1 2 Radio Radio is one of the most important boards on every satellite It provides com munication between Earth and downloading results or uploading configurations Transmission frequency is in radio amateur free band at 436 MHz with communica tion speed 9600 Baud University of West Bohemia is responsible for communication with the sat They will download data and
61. ocess picks up two points from memory performs FFT algorithm and stores new points on the original positions After that program finds the highest peak which corresponds with resonant frequency 15V 15 15V 6 E Switch to generate E zi damping envelope C ARTICLE FROM JAPAN Store results into memory for further dispatch to the Earth Results of frequency are stored into internal EEPROM memory of microcontroller for further dispatch to the Earth Memory data which corresponds 40 hours of recording and hold data also after power loss attenuation and resonance can include amount V FUNCTIONALITY VERIFICATION During writing a program for development kit was necessary to check all functions It leads to create device which one generates approximately the same signal as it gets from oscillations on the probe With this idea came professor Yana who provided us draft of schematic which is redrawn by Eagle software illustrated in the picture fig 8 The circuit produces oscillations with exponential envelope To the circuit must be connected external oscillator as input of the board The board consists of some parts as monostable gate output is connected to analog multiplier through capacitor which produces exponential envelope To the multiplier is also brought external oscillator Both inputs are multiplied in integrated circuit AD633 Output is connected to adder with another inpu
62. oducibility of measurement in the same conditions All measurements precede theoretical calculations and simulations of gained signal natural freguencies of cantilever etc and then we compare real data to theoretical results At first it is good to know natural freguencies of oscillated cantilever From eguation 4 11 natural freguencies are gained For HM panel with these parameters e cantilever dimensions 13 2 x 67 4 x 1 2mm e elastic modulus is E 3 8 10 Pa e material density is p 2300kg m the first three natural frequencies are 171 6 Hz 1075 Hz 3011 Hz The first two frequencies satisfy Shanon Niquist theorem that says the sampling frequency fs 4kHz must be at least two times higher then maximum sensed frequency Third and next cannot be directly measured 100 lt lt gt 80 2 a o 2 2 o o N N S 20 0 0 0 500 1000 1500 2000 0 500 1000 1500 2000 Freguency Hz Freguency Hz a Spectrum of panel mounted in clamp b HM panel mounted on a frame and all hanged on wire Fig 8 1 Difference in spectra with different placement Pictures of real measured signal in Fig 8 1 are comparing the difference of spec trum in dependence on placement of HM panel Both charts contain first two natu ral freguencies but the second one Fig 8 1b has extra peaks on freguencies around 300 Hz and 750 Hz Peaks are probably caused by other resonances of a metal
63. on composite on nanosatellite miniCube mission QB50 Instructions 1 Study the problematic of evaluating the attenuation of exponentially attenuated signal generated by oscillations of carbon composite 2 Implement an algorithm to evaluate the attenuation of the measured signal 3 Study the suitable methods of excitation of the carbon composite and sensing its oscillations 4 Design a system for measurement of mechanical changes of the carbon composite in dependence on the temperature change 5 Implement a suitable wiring from 3 and 4 for the Payload of nanosattelite miniCube mission QB50 6 Implement the measurement under the RTOS system 7 Check and evaluate the functionality of the system References 1 JAN J slicov filtrace anal za a restaurace sign l 2nd ed Brno VUTIUM 2002 427 p ISBN 80 214 2911 9 2 T MA J Zpracov n sign l z skan ch z mechanick ch syst m u it m FFT Praha Sd lovac technika 2000 168 p ISBN 80 901936 1 7 3 HANA P INNEMAN A DANIEL V et al Mechanical properties of Carbon Fiber 3 Composites for applications in space Proc SPIE 9442 Optics and Measurement Conference 2014 2015 no 1 DOI 10 1117 12 2175925 Supervisor Ing Ladislav Sieger CSc Assignment validity 31 8 2016 L S prof Ing Miroslav Hus k CSc prof Ing Pavel Ripka CSc Head of department Dean In Prague 16 2 2015 ABSTRACT This master s thesis talking about mea
64. ould sample signal up to two million samples per second So it leads to averaging four points groups to eliminate glitches All data are stored into SPI SRAM memory from which are further loaded Computing damping envelope First what is needed to do is to perform mathematically absolute value to get one sided envelope Next step is logarithming signal because it is needed to get attenuation which represents b in exponential expression as shown in formula 2 Then is applied moving averages to smooth signal and computing directive using least squares method 2 60 m 55 2 50 gt 45 2 a 4 E lt 5 30 0 0 2 0 4 0 6 0 8 4 Time 5 Fig 6 Attenuation computed by least squares 10000 Fourier amplitude 5 E 5 3 0 100 39 200 Frequency Hz Fig 7 FFT result by development kit Computing Fast Fourier Transform It is known that resonance frequency of new material is in range 100 200 Hz so the signal sampled by 8 kHz must be decimated to 500 Hz to observe condition of the highest accuracy in frequency FFT algorithm decimation in time is calculated in two steps First step is to swap sampled points according to address which is bit reversed and next step is calculating Fast Fourier Transform Program performs decimation and bit reversing simultaneously Program chooses one point and stores it to the bit reversed address into SPI memory Right address positions are stored in the table Computation pr
65. pment kits with this MCU Picture of it is in the photo Fig 8 3 The kit consists of main MCU with three user buttons alphanumeric Liquid crys tal display LCD two LEDs and Universal synchronous and asynchronous serial receiver and transmitter USART which could be used for bootloader for down loading a new version of FW into the chip The kit has the advantage during debugging in that display could serve feedback what is happening in the MCU USART can be advantageously used for downloading measured data or uploading commands for affecting processes in the chip 48 75 8 2 Research During Internship in Japan 6 Fig 8 3 Photo of development kit 24 One of the first things what were needed to do was commissioning ADC and downloading signal via USART ADC was set up as single ended input on first analog input of MCU with a result saving into internal SRAM Amount of samples which could be saved is estimated at 2500 points in 12 bits resolution 16 bits variables These points are downloaded into PC and then evaluated For evaluation of signal in the computer environment Matlab was used Next step was implementing the FFT based on sampled and evaluated signal 8 2 Research During Internship in Japan Part of the research I and my colleagues Martin Urban and Veronika Stehl kov did was done during summer internship in 2014 at Hosei University in Tokyo Japan There has been done implementing of FFT into chip together wi
66. readboard or testing board works a printed circuit board proposal follows Last task is to assemble the final version last testing and then the device can come to mass production or to science use as in case of our probe II MAIN GOALS OF OUR EXPERIMENT Whole our work was divided into several parts following one after another First we had to design printed circuit board for main measuring This is first step because one testing board must be produced and proved in many tests before the launch At the moment when design was completed we can start design and write a program for measuring changes in the resonant frequency due to composite aging The main goal is to create and write a program for microcontroller which will measure signal with unknown frequency in range 100 200 Hz composite material s resonance frequency by Fast Fourier Transform with the highest possible accuracy in frequency and resolution approximately 0 1 Hz Another purpose is C ARTICLE FROM JAPAN calculating exponential envelope damping factor Microcontroller should be able to communicate with on board computer OBC via PC interface Next goal is measuring temperature by PC thermometer radiation and evaporation Measuring evaporation other gas from the material will ensure by five humidity sensors These sensors are not only sensitive to humidity but also on some other gases which may be released from the material To check the will be placed several
67. rt of master It is commonly used for ACK or returns data to the master from slave Last five bits are configurations of transmission and all devices on the bus must have same settings of these bits 34 75 o 6 3 FreeRTOS After header follows data Maximum size of data is not generally set but it is not recommended to transfer huge amount of data because transfer could overload bus and other boards have to wait until bus is free It has been established that optimal size of data is up to 64 Bytes per packet Disadvantage of sending huge amount of data is memory demands on both sides of communication and busy line of one transfer Whole CSP is suited for running on FreeRTOS 21 6 3 FreeRTOS FreeRTOS is shortcut for Free Real Time Operating System This system is multi platform and it can run on 32 bit platform as well as on 8 bit It is primarily suited for embedded systems based on ARM cores but it is possible to use on small devices such as ATxMega Only requirement of FreeRT OS is one dedicated system timer Timer is used for switching multitasking of each used task which are declared in the beginning of the code Recommended time for switching tasks is 1 ms When shorter time is used it has been observed that MCU has been halted by switching between tasks m 6 3 1 Creating Tasks Every task has some necessities when it is created task name and function place where it is created memory allocation size task priority
68. s Temperature of HM Board In case any board needs to know the temperature from HM board it can send a packet to this port It is sufficient to send any packet to the port 21 and HM board responds by defined message Code 6 2 37 75 6 COMMUNICATION WITH OTHER BOARDS Code 6 2 Respond to Get Temperature Command packed uint16_t temperature temperature from ADT7420 typedef struct __attribute_ hm_temperature_t 6 5 Data Keeper Data Keeper DK is used for storing data such as results of measurements log files etc Data are saved on Secure digital SD card which is formatted by UFFS file system suitable especially for embedded systems NAND flash memories 23 Every result is stored into DK and each board has a few containers for storing data Containers are separated files sorted by CSP address of board and a vari able storage number All data must be divided into chunks each with unique identification number Each storage when it is created has fixed data space for configuration data chunk number zero When chunk zero is used for configuration data it must to be created before the first use a storage Measurement Process in HM board Before the measurement starts it is necessary to send activation sequence to the board It is provided by OBC through CSP command Header of CSP contains destination port which correspond to specific command Port chosen for the main measure
69. sampled signal by using FFT It is supposed that the cantilever produces damped oscillations with exponential envelope and signal is then analysed by linear regression All these processes are discussed on the following pages 4 1 Damped Oscillations Damped oscillator is in the illustration Fig 4 1 with spring which is represented by Spring constant k weight at the end by Mass m and deviation by z N N m Fig 4 1 Picture of elementary damped oscillator These oscillations could be described using following equation 4 1 F Fa F 0 4 1 where d x Fy b 4 3 IT Tat en 4 MECHANICAL CHANGES MEASUREMENT A 5 o and meaning Inertial force F Damping force Fa Reverse force When equations 4 2 4 4 are put into 4 1 it gets differential equation of second order 4 5 d r dx gt 4 ma n 4 5 Solving it gets equation for damped oscillations a t 207 sin re 4 6 If it is known that w w 02 formula 4 6 could be rewritten into the form for damped sine oscillations 4 7 a t zoe sin wt 4 7 where is Attenuation of system w is Angular frequency and x is Deviation 0 5 Amplitude 0 0 05 0 1 0 15 0 2 Time 5 Fig 4 2 Simulated damped signal o 4 3 Cantilever Oscillations 4 2 Elementary Oscillator Elementary oscillator could be described on simplified example as a strin
70. se of more than 120 measurements results could be same Amplitude of signal at beginning and at the end of measurement is the same This could be caused probably by changes of material after fabrication This claim was not possible to verify during the finishing of this thesis because another measurements have to be performed 3 1 25 T 3 21 124 5 z S a E lt 3 3 235 attenuation temperature 3 4 i i i 22 0 20 40 60 80 100 120 Samples Fig 8 10 Chart of temperature and attenuation in time 55 75 8 MEASUREMENTS TESTING AND RESULTS o o 8 7 Conclusion Many measurements were made For example the tests in the oven included frequency and attenuation change depending on the temperature Results of them were surprising The value of natural frequencies do not change in dependence on the temperature Value of attenuation in this case has changed at temperatures above 80 C according to Fig 8 6 For trustworthy results more measurements at one stable temperature should have been made than one as seen in the Fig 8 10 These measurements are very time demanding at least waiting to the right tem perature another problem is computing time of signal processing in the chip and also the download of signal of each measurement This time was estimated to 27s per one measurement estimation is based on previous measurements of 120 samples Also it would be goo
71. t from potentiometer which provide DC offset Output of the board should be DC shifted in range 15V The board has power supply 5V and converted to 15V through DC DC converter Output voltage damped oscillations depends on voltage of oscillator and envelope 15V gt gt i ie Input signal 5 le 18V 3 GND 7 3 1 r 16 3 E 4 GND GND Fig 8 Damping exponential envelope generator schematic ARTICLE FROM JAPAN Final board is in the picture fig 9 Professor Yana was benevolent and arranged produce at manufacturer who create some pieces of boards which we assembled from parts who gave us We are very pleased to create and then testing on it damping oscillations measured by development kit Fig 9 Damping exponential envelope generator Printed circuit board The whole board was assembled and tested under supervision and technical assistance of Mr Saitoh and Mr Iki They kindly provided us their laboratory with all necessary equipment like oscilloscope signal generator or soldering station Final assembled board is in the picture fig 10 Fig 10 Assembled damping oscillator VI CONCLUSION During our internship we collaborated mainly with professors Kazuo Yana Gaku Minorikawa and Akira Yasuda These professors helped us to realize internship and during the whole
72. t in least square method s Amplitude point in least square method dB XXII 1 Introduction This master s thesis is generally talking about mission QB50 small nanosatellites called CubeSat Main part of the thesis is measuring mechanical changes of carbon fibre material in time It seems to be quite easy task but there are many aspects which could happened and all dangerous states that must be prevented So here is a reason to test all climatic conditions mainly in vacuum and operation temperatures on orbit The probe also has many other measurements such as X Ray measurement of the Sun called Medipix evaporation measurement verifying quality of the carbon fibre shielding and some other measurements For example issue of evaporation is detailed and described in the thesis Measurement of evaporation and evaluation of changes of the mechanical properties of carbon composite on nanosatellite miniCube mission QB50 1 of my colleague Bc Martin Urban Or the issue with radiation shielding has in charge my other colleague Bc Veronika Stehl kov and it is more detailed in her thesis Radiation resistance measurement on nanosatellite miniCube mission QB50 2 Tested carbon fibre material could be used on a new satellites in future and it should replace old and heavy tungsten shielding It would lead to decrease in weight of the whole satellite and launch cost The device for measuring mechanical changes in time consists of excitation
73. t of Acronyms XIX List of Symbols XXI 1 Introduction 1 2 Mission QB50 3 2 1 One CubeSat unit 3 3 CubeSat VZLUsat 1 3 1 Parts of Probe 3 1 1 X Ray Optics and Medipix 6 3 12 Radio 7 3 1 3 Scientific Unit 7 3 1 4 Volatiles Board 7 3 1 5 Radiation Shielding Measurement 3 1 6 HM System 3 1 7 On Board Computer 8 3 1 8 Electronic Power System 9 3 1 9 Stabilising system 9 4 Mechanical Changes Measurement 11 4 1 Damped Oscillations 11 4 2 Elementary Oscillator 13 4 3 Cantilever Oscillations 13 4 4 Measuring by Accelerometer 15 4 5 Measuring by Piezo 15 XI 46 How to Excite the Cantilever 16 4 7 HM Panel x 16 4 8 Measured Parameters 16 4 8 1 Young s Modulus of Elasticity 17 4 8 2 Attenuation of Signal 17 4 9 Computing Process of PRI 4 amp 2 42 4 18 4 10 Computing Process of Attenuation 20 4 10 1 Rectify and Moving Averages 21 4 10 2 Logarithm Moving Averages 22 4 10 3 Least Square 22 Device for Mechanical Changes Measurement 25 5 1 HMpane 25 5 2 HM board lt as 5 5 27 5 2 1 Microcontroller for Payloads 27 5 2 2 Oscillator 28 5 2 3 External Memory 29 5 2 4 Power Switch 29 5 2 5 Piezo Connection 30 5 2 6 Thermometer 31 Communication with Other Boards 33 6 1 IC Interface 33 6 2 CubeSat Space Protocol 34 6 3 PreeR TOS e ses 35 6 3 1 Creating Tasks 35 6 4 Communication with HM Board 36
74. th storing data to the external SPI memory Knowledge about signal processing was provided by professor Kazuo Yana Ph D to whom I would like to express huge thanks 25 as well to other staff of the University for providing laboratories and equipment for testing and verifying progress of implementing functions into the MCU First step of verification was done on sinusoidal signal with triangular envelope as shown in the Fig 8 4 The implemented FFT was calculated with 256 points only time saving for verifying that the algorithm works with external memory Results were what had been expected More points of FFT calculation were added later to one of the last versions of Firmware FW or for the calibration measurement Second step was verification of signal attenuation Basics could be performed on triangular envelope but to get closer to real situation it is better to measure 49 75 8 MEASUREMENTS TESTING AND RESULTS o o o o 95 1000 lt 100 o 500 S A 2 0 a 50 gt 500 2 O 1000 S 0 ae 0 0 5 1 1 5 0 100 200 300 400 500 t s Freguency Hz a Sine signal with triangular envelope b Spectrum of signal with using FIR filter Fig 8 4 Signal with triangular envelope a signal with exponential envelope For testing purposes a generator was made at Hosei University which produced the required waveform The generator needs only a power supply and an
75. th wires would cause bigger attenuation and frequency change Results of measured cantilever would be changed there would be a big deviation from theoretical results 10 4 5 Measuring by Piezo Another method how to measure oscillations is by piezoelectric plate Here are some reguirements for placing it for higher sensitivity and it is the place with the mechanical stress of the measured material and that position is at a fixed end of 15 75 4 MECHANICAL CHANGES MEASUREMENT o cantilever When piezo is stressed it produces electrical voltage which is measured it corresponds to mechanical oscillations in this case 10 4 6 How to Excite the Cantilever Here is another issue and it is how to excite the cantilever One of the possibil ities is to tap it by finger but it is not suitable for autonomous measurement or in space There are some other possibilities and one of them is to mechanically tighten the cantilever and then let it go This situation is not reliable from my point of view There should be a precise mechanism There is another method and it is electrical excitation of beam At the opposite side of the cantilever is a fixed coil which attracts the beam by electric pulse to the coil The question is how to attract it One possibility is a glued permanent magnet Permanent magnet has some problems It is quite heavy and it will affect the measurement So there is a material called permalloy which is a material
76. tional radiation from inside from irradiated construction of probe there will be not only measured signal from space but this unwanted radiation too It is necessary to try to separate both environments or subtract it for relevant results Space cube carries also many other tasks For example there will be sensors for measuring temperature in different parts of probe or humidity sensors made by Innovative Sensor Technology s r o which will measure vaporization rate of tested composite during transition from atmosphere to vacuum These tasks don t have so high importance and due to the power limitation of whole device they run only when main tasks are switched off like in case of Timepix doesn t look into the Sun That means all tasks are sharing processor time as in time multiplex and have solar and backup battery power together X ray optic Deployable So lar panel Measuring mmn board XRB diodes Antenna Volatiles X ray detector Radio OBC PWR SCU Fig 2 Measuring boards and inner arrangement Measured data will be sent to the headquarters in Pilsen when the probe will pass over Czech Republic It will be approximately twice to day The speed of data transmission will be changing as well as the time go Data amount depends on where the probe will be and how long can communicate according to height o
77. to the result of moving averages we should get a line Red line in the picture Fig 4 10 represents the result of logarithm and the outcome is what was expected a line 0 01 02 03 04 05 06 07 08 09 1 t s 0 ji L L L Fig 4 10 Attenuation with directive 4 10 3 Least Square Last step of evaluation of attenuation is compute directive of line Here are two possibilities First is defining two points and calculate directive from them Second 22 75 o 4 10 Computing Process of Attenuation 6 more sophisticated and more computing expensive is using an approximation of curve but more precise than using method of two points The most elementary is linear approximation called least square method or linear regression Also waveform could be better approximated by a higher degree of regression or another type such as exponential or logarithm For our purpose linear regression is enough and its representation is equation of line for example as y ka q where Offset of line k is computed by equation 4 27 and Offset of line q is computed by 4 28 k NY 2698 Do HDi 4 27 NEG gt BD zj q Kii A yi Yi Ti D Tiyi 4 28 Ny z baal r where N is Number of point least square method x Time point in least square method y Amplitude point in least square method k Directive of line q Offset of line Linear regression could be used for the whole signal only in the
78. tor while n lt NO_POINTS number of iterations log2 NO_POINTS pointer_A MEM_FFT_BEGIN address of beginnig memory space for FFT pointer B pointer A n 8 Fnk 0 for k 0 k lt n k distance between butterflies dsin read_sincos Fnk read sin and cos from table dcos read_sincos Fnk for s 0 s lt angf s 4 butterflies A spi_mem_read_complex pointer_A spi_mem_read_complex pointer_B real B real dcos B imag dsin imag B real dsin B imag dcos real A real C real imag A imag C imag real A real C real gt e U U QA QA U imag A imag C imag spi_mem_write_complex pointer_A A spi_mem_write_complex pointer_B B pointer_A n 16 nextpoint in memory 2x 2x4B pointer B n 16 pointer A NO_POINTS 1 8 pointer B NO_POINTS 1 8 Fnk angf 1 2 n lt lt 1 angf gt gt 1 twiddle factor 2 7 MEASUREMENT PROCESS IN HM BOARD o o 7 5 Signal Processing Attenuation After computing FF T the next step is to calculate attenuation of signal which is expected to has exponential envelope By using equation 4 27 the directive is computed attenuation of signal from linearised signal There is a dedicated memory space for the whole computing process of directive and has same size as sampled signal Sample of code is in the listing Code 7 4 where is only the main part of calculation Code 7 4 Calculation of attenuation
79. upload configurations to the probe And also process some data from it Communication board which will be launched is already tested with antenna and radio board and works fine on the ground in 10 km distance 6 E 3 1 3 Scientific Unit Every nanosatellite should have one of three scientific units Tab 3 1 Scientific units 3 Type Description INMS lon Neutral Mass Spectrometer FIPEX Flux Probe Experiment m NLP multi Needle Langmuir Probe All scientific units are equipped with a thermometer such as Thermistor Ther mocouple VZLUsat 1 has FIPEX as scientific unit The unit measures behaviour of at mospheric oxygen at lower thermosphere It is important for exploring erosion on surfaces of spaceships during contact of atomic oxygen with surface of probes 3 7 E 3 1 4 Volatiles Board Volatiles board is assembled with humidity sensors which are looking for residual humidity or evaporation from the whole probe mainly from carbon fibre materials It has three types of sensors two from JST company labeled HYT271 and HY T9329 both in two of each Third type is HAL2 in three pieces from TTS company It is not only sensitive to humidity but a little to other gases All sensors are connected to the board with a driver called PicoCap which is converting capacitance into ratio compared to the reference capacitance All HYT sensors and PicoCap communicate 7 75 3 CUBESAT VZLUSAT 1 o via 12C bus con
80. was divided into several functions depending on called function computing attenuation FFT etc Further testing was performed on OBC emulator which runs under Linux and communicates via serial port through development kit where the data were trans lated into CSP packet and sent via I C bus to the device in this case HM board This solution has the advantage that it is not necessary to have flight board of OBC for testing There are some disadvantages as well such as testing on real HW or communication with other boards such as measure board The implementation of functions to the chip of HM board was made according to the chapter 6 4 where tasks in dependence on communications ports are simply specified Also it was necessary to write code into OBC for communication with HM board The code contains only start commands of specified actions such as start measure process get raw signal and etc Data which can be saved are send directly from the board to the datakeeper and stored Some measurements were performed and worked fine for example sampling signal receive data temperature or final results When the program is written it is recommended to transfer data in network format big or little endian to avoid a conflict with format Then when data should be used they must be decoded from network into board format This could be a source of errors when the data are decoded 8 6 Final Measurements Before Flight Before the final
81. which if it pass could be used as shielding material for standard satellites e g GPS and communication satellites etc in the future Naturally due to the rising popularity of this cheap nanosatellites like type QB50 this proved material could be used for producing standardized parts of skeleton and shielding for them too It means an advantage for future experimentators who will be able to fully concentrate to their research itself without solving questions of construction and shielding Wide angle X ray imaging system with Timepix detector is a special type of lens which works with reflexivity instead of refractivity Because the probe doesn t have active engine there are at least coils for all three dimensions which will slightly orient the probe in dependence on Earth s magnetic field One of next parts of measuring is measuring of humidity There will be several sensors in the probe which are connected with the main computer through PC The computer has to be programmed to switch between desired sensors and get measured values Last but not least of our tasks is measuring of space radiation and shielding capabilities of composite panel The probe will have three measuring diodes aboard and they will be shielded by none one and two layers of composite This task is complicated due to low sensitivity of diodes in case of high temperatures and cooling in free space is hard to solve There also is a problem with addi
82. z and minimum fx min have to be less then maximal value fx max In the case this condition is false minimum and maximum value will be swapped In case freguency exceeds limits of half decimation freguency maximum value is set this value Maximal freguency which can be set is 65 kHz enough for this application Time for power switch on time power switch on is in 1 10 of ms number 15 corresponds to 1 5 ms The maximum time is 25 5 ms 255 Decimation factor decimation factor is a value for calculation of spectrum from sampled signal This value is division factor for decimation of sampled signal to reduce spectrum of signal The directive of exponential envelope is calculated by the least sguare method length of 18 is constant which specifies computing range of this method Value represents number of calculation points of linear regression 7 2 Get Temperature Orientation and Time Health Monitoring panel has six PT1000 thermometers in total which measure thermal transfer through material and or surface of it Temperatures are available on Measure board and they can be obtained by CSP command The board return temperatures and MCU on HM board store them to the memory for further transfer into DK Orientation and board time is available on OBC Process of reading is same as temperature but with other board 40 75 7 4 Signal Processing FFT 6 7 3 Excite of Coil and Sampling of Signal When all parameters are lo

Measurement of changing mechanical properties of carbon

Contents

Download Pdf Manuals

Related Search

Related Contents