Home

Sin título

Contents

1. chan The number of channels fd file Th handle of the file where to store the data Returned gt 0 The number of errors encountered here kkk k k k k k ok k k k k k k k k k k k k k k k k k k k k k k k k k 65 int save data profiler FILE fd file int chan x 832 buffer long samples int i long 1 long samples sizeof buffer CHANS ENABLED int errs 0 printf Number of samples per buffer 14 samples gsc label Saving data n for 1 0 1 lt samples 1 1 Only chO and ch3 are included on the file to make it of a reasonable size if 1 4 chan 0 110 4 3 0 t i fprintf fd file 081X long buffer 1 if i 1 10 1 printf FAIL fprintf failure n errs 1 break if 1 4 chan chan 1 1 i fprintf fd file r n yf 2 66 printf FAIL lt fprintf failure errs 1 break return errs Paak k k kkk kkk k kkk k kkk k kk k KKK k K K Function _channels_profiler Purpose Check to see how many channels the board has Arguments fd The handle for the board to access chan Report the number of channels here Returned 67 gt 0 The number of errors encounterred k k k k k k k k k k k k k k k k k k s32x chans gt int channels pr
2. km mama paa Ware aay eel Entrainment done 4 5 4 NI y Y Convective 4 Boundary Layer gt Residual layer K a k Tr x gt E t i intermittent turbulence i r 1 i L i D y 4 CAA Noctumal ME 4 warm ur TE EYA Boundary Layer EAR MOL is L Surface layer j fre tM Jow keveljes Time Afternoon Sunset Midnight Figure 2 1 Boundary layer structure from 6 2 2 Clear Air Backscatter Theory Backscattering occurs from refractive index irregularities in clear air and has a relationship with the atmospheric structure and turbulence as determined from ex perimental studies of angel echoes starting with Plank 11956 and Atlas 1959 Radar backscattering from refractive index variation is able to provide helpful information about the atmospheric structure First radar outlines regions of increased refractive index variability because of the enhanced backscattering Second the radar backscatter contains quantitative information about the variability in the refractive index field The radar backscattering from the clear air atmosphere is caused by irregular small scale fluctuations in the radio refractive index produced by turbulent mixing and is dependent on the atmospheric water vapor temperature and pressure The intensity of these fluctuations can be described by the structure constant C In 1969 Ottersen 10 was the first to derive t
3. k k k Function _parse_args Purpose Parse the command line arguments Arguments argc The number of command line arguments given argv The list of command line arguments given Returned gt 0 The number of errors encounterred RK SR k k k k ok k k ok k k k k k k k k k ok k k k k k k k k k k k k k k k k static int parse args int argc char argv 1 char c 44 int errs 0 int i int j int k printf USAGE savedata lt c gt lt C gt lt m gt lt n gt lt index gt n printf C Continue testing until an error occurs n printf C Continue testing even if errors occur n printf m Run for at most minutes a decimal number Nn printf index The index of the board to access n printf NOTE Hit Ctrl C to abort continuous testing n for i 1 i lt argc i if strcmp argv i c _continuous 1 _ignore_errors 0 continue if strcmp argv i C _continuous 1 ignore errors 1 continue j 45 if argv i 0 amp amp argvlil 1 m amp amp argvlil121 1 j sscanf amp argv i 2 AdAc amp k amp c if j 1 amp amp k gt 0 _minute_limit k continue errs 1 printf ERROR invalid argument s n argvlil break j sscanf argvlil Ad
4. k I k k k k k kK k k k k k k k k int main int argc char argv 1 int errs time t exec time NULL int fd 0 FILE fd file long hours long mins time t nov int qty int ret EXIT FAILURE long secs time t t limit time t test 49 char datebuf 128 char filename 128 struct tm today time t starttime __s32 chans 32 int got gsc label init 26 test time NULL printf savedata Capture and Save Data to Disk By David Garrido Version s n VERSION errs parse args argc argv gsc id host Identify the host os and machine t limit exec minute limit x 60 qty gsc count boards DSI BASE NAME Count the number of installed SI04 errs gsc select 1 board qty amp index Select the board to utilize when more than one is present if qty c 0 errs gsc_label Accessing Board Index printf 4dNn index 50 fd gsc dev open index DSI BASE NAME Perform an open on the device with the specified index if fd 1 4 1 printf ERROR Unable to access device 4 index if errs 0 ret EXIT_SUCCESS errs perform tests fd V V MM MM MM P MP M PM MIMPMI gsc_dev_close _index fd test finished so we close now time NULL secs now test hours secs 3600 secs secs 2 3600 mins secs 60 secs secs 60 printf duration 41d
5. 1d 021d n hours mins secs printf Test done n 51 if ignore errors 0 amp amp errs errs if minute limit amp amp nov gt t limit errs We finish testing and we start adqiring data if lerrs _ignore_errors 0 1 printf Starting adquisition I 11 ITI EIN fd gsc dev open index DSI BASE NAME dsi config board fd 1 1 SAMPLE RATE errs channels profiler fd amp chans Creating the file printf Creating the file where the data will be stored n time amp starttime today localtime amp starttime strftime datebuf 128 Y m44 H2M S today sprintf filename s dat datebuf fd file fopen filename w b if fd file 1 errs 1 else printf file created n 52 for 1 now time NULL errs 0 got dsi dsl read fd buffer data sizeof buffer data CHANS ENABLED if got O 1 errs 1 else if got sizeof _buffer_data CHANS_ENABLED errs 1 printf FAIL lt got ld samples requested ld n long got long sizeof _buffer_data CHANS_ENABLED else errs 0 printf PASS 714 samples n long sizeof _buffer_data CHANS_ENABLED errs _save_data_profiler fd_file chans _buffer_data sizeof buffer data CHANS ENABLED if minute limit amp amp now gt t limit break 53 if ignore errors 0 amp amp errs break
6. de DATAAGCOUTSEETTON ys cr pad ck we A 25 4 1 24DSI6LN Data Acquisition Board 26 4 2 Data Acquisition Program pa ev Sues PANAY YANG cesa c au sere 27 43 Goynclromzal sn Programe erpa Pete Dex ane ia Sea 28 r SIGNAL PROCESSING LU a saa la Dee PESCE em cS 30 SU Signal Processing rota eyer celts dade cot eut k sl dee n eto so 30 5 2 Noise Floor Hatimatiot pada Na KAKA DA E ANG PIG peed one MTS 31 5 3 Minimum Detectable signal iss Less 92 6 ANTENNA ISOLATION cot stings eel eee aeee 35 6 1 Antenna Isolation Tests With New Shrouds 36 T GRAPHICAL USER INTERFACE 39 S UOMCLLLO XOL Ana scuba died se s Red Sacs eeu ed aud 41 GOL EH ii ds Le an oba te TN EE AD s 41 Baa BULUT Work aa la ala ace aeons MN Id ea x z l 42 APPENDICES A DATA ACQUISITION PROGRAMS 43 AT a NAAN que s oa die S d 43 2 27 ACV AS ES ERE AN AA 54 B SYNCHRONIZATION PROGRAM 69 B l pre processing david pro 34 04 eo oom RETIRER se ot Sie ione ALAGA 69 vi C SIGNAL PROCESSING PROGRAM 74 C 1 enmdprof process david pred a ad de EE Rp See en AA AR 74 D GRAPHICAL USER INTERFACE PROGRAM 80 DIL Wwe mew create DEOS ee ET rco qr aug ERA ORE S ALCUNE d 80 D2 ana event A S e PR DS BERE CRETAE A ated 84 Da getconf itl
7. widget label bi value widget base b0 row widget label b1 value Number of channels widget text bi1 editable xsize 8 uname channels value 3 widget label b1 value channels 82 Radar buttons bi widget base base frame column 50 widget label b1 value Radar configuration b2 widget button b1 value Radar Start uname radar start uvalue radar start b2 widget button b1 value Radar Stop uname radar stop uvalue radar stop t2 widget label b1 value Data collection b2 widget button b1 value Start Data Collection uname data start uvalue data start b2 widget button b1 value Stop Data Collection uname data stop uvalue data stop b1 widget base base frame column t2 widget label b1 value Display Data b2 widget base b1 row t2 widget label b2 value Time Average t2 widget text b2 editable xsize 8 uname time_avg value 61 t2 widget label b2 value s 83 b2 widget button b1 value Display Data uname display data uvalue display data setenv INIT 1 widget control base realize xmanager wgui_fmcw base END D 2 wgui fmcw event pro David Garrido Lopez MIRSL Lab UMass Amherst GUI For FMCW Boundary Layer Radar PRO wgui fmcw event ev COMMON SHARE1 hdr display d
8. 2 Chirp Length ms chirp time id widget info ev top find by uname prf time widget control id get value prf time printf 1 PRF Hz sprf tine printf 2 PRF Hz 2 pri_time id widget info ev top find by freq widget control id get value A0 freq printf 1 AO Sample Frequeny kHz freq printf 2 AO Sample Frequeny kHz A0 freq id widget info ev top find by uname samp freq widget control id get value samp freq printf 1 Sample Frequency kHz freq printf 2 Sample Frequency kHz samp freq id widget info ev top find by uname scans profile widget control id get value scans profile printf 1 Scans per Profile scans profile printf 2 Scans per Profile scans profile id widget info ev top find_by_uname profile 8T widget control id get value profile printf 1 Total Profiles profile printf 2 Total Profiles profile id widget info ev top find by uname scans frame widget control id get value scans frame printf 1 Scans Per Frame scans_frame printf 2 Scans Per Frame scans_frame id widget info ev top find_by_uname channels widget_control id get_value channels printf 1 Number of Channels channels printf 2 Number of Channels channels close 1 close 2 end D 4 run vindprofiler pro pro run vindprofiler 3 Compile and execute windprofiler c that init
9. 3 1 VVind Profiler old configuration 12 3 1 2 FMCW Radar Previous Results After testing and deploying the FMCW radar it turned out that there were some noise and sensitivity problems in the previous design that needed to be fixed In this section the main issues that impede detection and data collection from the FMCW radar will be explained in a consistent way and possible solutions will be presented Audio Amplifiers The main problem with the audio amplifier used in the old design is due to the noise it generates itself Looking through the specifications it was found that there were about 50V rms of noise introduced at the input of the amplifier such as shown in Figure 3 2 Due to the gain of the receiver it is necessary to make an audio filter amplifier with low noise components that do not increase the noise figure of our receiver that Input noise G 40dB G 40dB much Figure 3 2 audio amplifier issue 13 Antennas The isolation of our antennas is not good enough since currently there is 71dB of isolation between transmitter and receiver With that isolation the amount of leakage received is of around 23 dBm which is a very high signal so it was necessary to attenuate 10dB in the front end of the receiver in order to not saturate the mixer That attenuation has bad effects on our SNR because we attenuate signal but not noise since the main noise comes from the audio amplifiers Therefore 10dB in
10. Bandwidth widget text b1 editable xsize 8 uname bw value 25 widget label b1 value Mhz widget base b0 row widget label bi value Chirp lenght 2 widget_text b1 editable xsize 8 uname chirp time value 8 3333 widget label bi value msegl widget base b0 row widget label b1 value PRF 2 widget text bi1 editable xsize 8 uname prf time value 100 widget label bi value Hz widget base b0 row widget label b1 value A0 sampling freq widget text bi editable xsize 8 uname A0 freq value 40 8l t2 bi t2 t2 t2 bi t2 t2 t2 bi t2 t2 t2 bi t2 t2 t2 bi t2 t2 t2 widget label b1 value Khz widget base b0 row widget label bi value Sample frequency widget text bi1 editable xsize 8 uname samp freq value 61 widget label b1 value Khz widget base b0 row widget label b1 value Scans per profile widget text bi editable xsize 8 uname scans profile value 610 widget label bi value widget base b0 row widget label bi value Total profile 2 widget text bi editable xsize 8 uname profile value 180000 widget label bi value widget base b0 row widget label b1 value Scans per frame 7 widget text bi editable xsize 8 uname scans frame value 30500
11. SNR were lost by the attenuation Figure 3 3 Shroud fence built around the Wind Profiler antennas In an attempt to try to improve the isolation between the antennas and reduce the leakage received shrouds for the antennas were developed in order to reduce coupling The fence started at the ground level and reached 15 to 20 cm above the antennas The two antennas with shroud fence are depicted on Figure 3 3 The Wind Profiler was tested again with the following results The antenna isolation improved by 2 dB but the improvement in clutter rejection was insignificant That suggested that the main path for the clutter is not the backlobe but the antenna sidelobes A new redesign of the antennas or the shroud can effectively address this issue 14 Vector Modulator The mission of the vector modulator was to try to cancel the leakage The Vector Modulator is a device capable of changing the attenuation and the phase of its input signal With that theoretically it should be possible to generate an opposite signal of the leakage received But once used in the radar the cancellation achieved was fewer than 13dB which was not enough Also the cancellation achieved was not the same throughout the entire bandwidth as it would be desired The leakage power and the best achieved cancellation are shown in Figure 3 4 Ref 0 dBm Peak Log 10 dB 1 Marker e oi zi 305 500000 Mz Mu 39 18 dBn V ayi A mol Mv Cent
12. and execute the command start To verify that the device has been loaded ex ecute Ismod The module name 24dsi should be included in the output if driver was started properly To collect data the program needs to run while the radar is operating so the programs that start the radar have to be executed before 4 3 Synchronization Program It is necessary to synchronize the collected data in order to know when a profile starts and stops to allow proper processing of the data in order to extract range and doppler information from it As explained before the synchronization via hardware is still being developed So in the meantime synchronization via software is done To do so an IDL program called pre_processing david pro was created for that purpose The main file pre processing david pro is provided and commented in Appendix B This program also uses other files not included in this document such as readcol pro The structure of the program is as follows First of all the file containing the data is opened and the data stored in columns is read and stored Then the data is 28 processed erasing the header bits from every sample and converting the information from hexadecimal to voltage values After that a block size smaller than the number of samples per profiler is created and the synchronization of data starts To synchronize the data the synchronization signal is also sampled and so the program
13. begin x abs x if xant x le O then x xant x 0 1 else x xant x 0 1 endif powl plot index x xant x endfor svindov plot rawdata window 4 plot 10 alog10 pow1_plot 30 xtitle samples ytitle Power dBm 78 for index2 0L size powi 3 1 do begin if powl plot index2 lt pow1_noise index2 4 std then powl plot index2 1e 10 endfor oplot 10 alog10 pow1_plot 30 COLOR 1500 oplot 10xalog10 pov1 30 COLOR OO0FFFF x oplot 10 alog10 powl noise 30 COLOR FFFFOO x print std std at eof close lun close lun out close lun noise print File created end 79 APPENDIX D GRAPHICAL USER INTERFACE PROGRAM D 1 wgui fmcw create pro David Garrido Lopez MIRSL Lab UMass Amherst GUI For FMCW Boundary layer radar PRO wgui fmcw create COMMON SHARE1 hdr display datafile lun display 1 datafile fmcw conf hdr header_w base widget base title UMass FMCW Radar GUI v1 0 row Positioner settings 80 bo 50 bi t2 t2 t2 bi t2 t2 t2 bi t2 t2 t2 bi t2 t2 t2 bi t2 t2 widget base base frame column widget label b0 value Radar Settings widget base b0 row widget label bi value Center frequency widget text bi1 editable xsize 8 uname centr freq value 115 widget label b1 value Mhz widget base b0 row widget label bi value
14. clock at the desired f provided by the FPGA Now a high speed clock is needed to do so and the fastest clock the FPGA can provide is at 10Mhz i a NA ili ii in y TAR PREPARADA ik ik so Ki sk s imd am sn T Figure 4 1 24DSI6LN Data Acquisition Card 26 The voltage range is the same as before being 10 V The board also provides its own anti aliasing filtering which helps to relax the audio specifications even more The sampling clock can be generated internally being able to chose from 2KSPS to 200KSPS with an accuracy of 25 PPM The sampling of multiple channel groups can be pre synchronized through soft ware or each group can be synchronized to an external hardware clock input So in order to synchronize the DAQ with the transmitted chirps a synchronization signal generated in the FPGA can be used for that purpose An acquisition program was developed using that external synchronization input signal feature It was tested sim ulating software synchronization pulses and it worked properly But when the real synchronization input signal was used it was not recognized by the DAQ Currently that problem is still pending to be solved waiting for the DAQ developers to identify the board s problem To be able to test the system an acquisition program without using the external synchronization signal was developed In order to synchronize the data software synchronization was used while the hardware
15. gsc_dev_close _index fd fclose fd_file return ret A 2 savedata c Rev 1814 Date 2009 04 22 17 13 15 0500 Wed 22 Apr 2009 Program developed by David Garrido Lopez include lt ctype h gt include lt stdio h gt include lt stdlib h gt include lt string h gt include lt time h gt include main h defines CCR ROO E K K Kk k k k kkk define _1M 1024L 1024L define TRIGGER_RATE 100 define SAMPLE_RATE 60000L 54 define CHANS ENABLED 6 variables ARK static u32 buffer SAMPLE RATE CHANS ENABLED TRIGGER RATE E kk k ke k kkk k k Kk k OA k k k k k k ok k ok k k Function channels x Purpose Check to see how many channels the board has Arguments fd The handle for the board to access chan Report the number of channels here Returned gt 0 The number of errors encounterred R R SR k k k ok k k k k k k k ok k k k A k k k k k ok k k k k k 55 static int channels int fd s32x chans 1 3 int errs gsc_label Input Channels errs dsi query fd DSI QUERY CHANNEL QTY chans if errs 0 printf ld Channels n long chans 0 return errs PROOF k kkk k kkk kkk k kkk k kkk KK Function _read_data Purpose Read data into the buffer Arguments fd The hand
16. has been customized and the FPGA has been modified to enhance the radar performance having now a 10096 duty cycle chirp For more information about it refer to 2 The new Data Acquisition Board arrived and it was integrated in the radar Data acquisition programs were developed for it but there are synchronization issues that should be corrected in the future Signal processing programs were developed for the FMCW radar allowing the radar to be tested in the laboratory All information will be showed in the following sections 3 2 Wind Profiler Subsystems Figure 3 5 shows a simplified block diagram of the FMCW Wind Profiler Apart from this there are also some other necessary components like several attenuators to adapt the input power levels The picture Figure 3 6 show the radar hardware deployed in laboratory Including the Wind Profiler boxes computer audio module and instrumentation Figure 3 6 Radar deployment in the laboratory 18 3 2 1 Control and Transmit Subsystem The communication between the operator and the Wind Profiler is made through the serial interface between the computer and the Field Programable Gate Array Board Altera Cyclone II EP2C20 FPGA Figure 3 7 that generates all the radar signals required for its proper operation The operator stores the necessary parameters to configure the radar in a file conf which will be sent via serial interface to the FPGA That file and its parameters can be modif
17. powl noise maximum max powl noise mean2 1 noise 50 198 for avg frame index 0L total profiles prof avg 1 do begin big loop for frame index OL prof avg 1 do begin readu 1 rawdata rawdata rawdata f frame frame index rawdata ch endfor f frame f frame winfunc 76 c frame fft f frame dimension 1 0 Block size 2 1 c mean total c frame 2 prof avg c framei c frame complex rebin real part c mean Block size 2 prof avg rebin imaginary c mean Block size 2 prof avg pow total double c_frame conj c_frame 2 prof avg powi total double c framei conj c framei 2 prof avg autocorretotal c frame 1 prof avg 1 conj c frame 0 prof avg 2 2 prof avg 1 autocorr phase atan autocorr phase autocorri total c framel 1 prof avg 1 conj c framei 0 prof avg 21 2 prof avg 1 autocorr_phasel atan autocorr1 phase Get velocity vel autocorr phase prf lambda 4 pi veli autocorr phasei prf lambda 4 pi Get coherence rho abs autocorr pow rhoi abs autocorri powi powdb 10 alog10 pow T7 powdb1 10 alog10 pow1 writeu lun out pow vel rho c mean writeu lun out powl vel1 rhol c mean endfor print fstat lun out size powl size powl noise avg 28500 58400 Std mean noise sqrt noise avg pow1_plot pow1 xant 1e 6 for index 0L size_pow1 3 1 do begin x pow1 index pow1_noise index 0 80 if x 16 O then
18. 0 break file fopen name w b if file NULL printf FAIL lt unable to create s n name errs 1 break 59 for 1 1 samples 1 4 i fprintf file 081X long _buffer 11 if i 10 printf FAIL fprintf failure to s n name errs 1 break if A chan chan 1 i fprintf file r n if i 2 printf FAIL fprintf failure to s n name errs 1 break fclose file 60 if errs 0 printf PASS s Nn name break return errs PK kk k kkk ak OO k k k k K k ok 2K K k k k k k k K K K Function save data Purpose Configure the board then capture data to a file Arguments fd The handle for the board to access Returned gt 0 The number of errors encounterred BO OO K k k kkk kkk k k k k k ok k k k k k k ok k kkk 61 int save data int fd 4 632 chans 32 int errs 0 gsc label Capture amp Save printf Nn gsc label level incO errs errs errs errs errs errs ll dsi config board profiler fd 1 1 SAMPLE RATE _channels fd amp chans read data fd dsi ain buf overflow fd 1 DSI AIN BUF OVERFLOW TEST NULL dsi ain buf underflow fd 1 DSI AIN BUF UNDERFLOW TEST NULL _save_data fd chans er
19. 2 6 the first 8 bits are discarded they indicated the channel only ch3 strmid ch3 2 6 print Data read 3 Conversion from hexadecimal string to double size array size ch0 size array size array 1 ch info dblarr size array 70 ReadS ch3 ch info Format 2 ch sync dblarr size array ReadS ch sync Format Z ch0 ch3 Now data should be transformed to real Voltage values 21LSB equals the full escale 20V divided by 1 1921uV 24 bits LSB 20 0000000 2 0000000724 0000000 ch info ch infoxLSB 10 transforming integer to real voltage value ch sync ch sync L8B 10 ch sync max max ch sync ch sync nin min ch sync ch info min min ch info ch info max max ch info ch info mean mean ch info ch sync mean mean ch sync print ch sync ch sync max ch sync min ch sync mean 2 ch info ch info max ch info min ch info mean Synchronize file and write the blocks to the file openw lun out outfile open the file file descriptors threshold 1 65 71 1 65 volts are chosen sync signal ideally between O and 3 3V Block size double fs PRF 2 It is needed to do that authomatically due to the no synchronization or jitter the minimum block size will be fs PRF 1 to be sure I put 2 writeu lun out Block size 200 We indicate what size block the file will have so processing will be faster if ch sync 0 lt threshold then aux 0 else au
20. AN UHF FREQUENCY MODULATED CONTINUOUS WAVE WIND PROFILER DEVELOPMENT AND INITIAL RESULTS A Thesis Presented by DAVID GARRIDO LOPEZ Submitted to Universitat Politecnica de Catalunya Escola Tecnica Superior d Enginyeria de Telecomunicacio de Barcelona in fulfillment of the requirements for the degree of ENGINYERIA SUPERIOR DE TELECOMUNICACIO January 2010 Electrical and Computer Engineering AN UHF FREQUENCY MODULATED CONTINUOUS WAVE WIND PROFILER DEVELOPMENT AND INITIAL RESULTS A Thesis Presented by DAVID GARRIDO LOPEZ Approved as to style and content by Stephen J Frasier Professor Electrical and Computer Engineering To my mother and father for the struggles they have gone through in their lives and the support they have always given me ACKNOWLEDGMENTS First of all I would like to thank Dr Stephen Frasier for giving me the chance to be here at MIRSL and for all of the help he provided me I have really learn a lot from you Being here at MIRSL has been an excellent experience that helped me to improve my knowledge in some areas such as remote sensing antennas signal processing or electronics It also made me grow as an engineer and experience other ways of working and researching that I am sure will be useful in my future career Also I have to thank Jorge Salazar Jorge Trabal and Rafael Medina the support and help they gave me They were always willing to answer my questions and that was always
21. Ac amp k amp c if j 1 amp amp k gt 0 2 _index k continue else t errs 1 printf ERROR invalid board selection s n argv il 46 break return errs paaa k k k k k kkk k kkk k kkk k kk k kkk k kkk k kkk K K KK Function _perform_tests Purpose Perform the appropriate testing Arguments fd The handle for the board to access Returned ko gt 0 The number of errors encounterred BO kkk k kkk k kk k k kK k k k k k k k k k k k k k k k k Static int perform tests int fd 1 AT int errs 0 const char psz struct tm stm time t tt time amp tt stm localtime amp tt psz asctime stm gsc_label Performing Operation printf s psz errs gsc_id_driver fd DSI_BASE_NAME Identify the driver errs dsi_id_board fd 1 NULL Identify the board errs save data fd return errs PLLC TTT okeokck k k k k k k k k k k k k k k k k k k k k k k k k k k k k 27 Function main Purpose Control the overall flow of the application 48 Arguments argc The number of command line arguments argv The list of command line arguments Returned EXIT SUCCESS We tested a device EXIT FAILURE We didn t test a device R R SR k k k k k k k ok k k k k k k k k
22. E NULL dsi ain buf threshold fd index 128L 1024 NULL dsi sw sync mode fd index DSI SW SYNC MODE CLR BUF NULL dsi channel order fd index DSI CHANNEL ORDER SYNC NULL dsi data format fd index DSI DATA FORMAT OFF BIN NULL dsi data width fd index DSI DATA WIDTH 24 NULL dsi init mode fd index DSI INIT MODE INITIATOR NULL dsi external clock source fd index DSI EXT CLK SRC GRP O NULL dsi fref compute fd index O amp fref dsi channel group source all fd index DSI CH GRP SRC GEN A 1 dsi fsamp compute fd index 0 0 fref amp fsamp amp nvco amp nref amp nrate amp ndiv errs errs errs errs errs errs errs errs errs dsi rate gen x nvco all fd index nvco 1 dsi rate gen x nref all fd index nref 1 dsi rate gen x nrate all fd index nrate 1 dsi rate gen x ndiv all fd index ndiv 1 dsi fsamp report all fd index 1 amp fref dsi xcvr type fd index DSI XCVR TYPE LVDS NULL dsi auto calibrate fd index dsi ain buf clear at boundary fd index dsi ain buf overflow fd index DSI AIN BUF OVERFLOW CLEAR NULL 64 errs dsi ain buf underflow fd index DSI AIN BUF UNDERFLOW CLEAR NULL return errs Paak k kkk OR k kkk k k Kk K K K Function _save_data_profiler Purpose Save the read data to a file customized for te FMCW radar Arguments Buffer The data to store
23. PGA was modified too in order to achieve a 10096 duty cycle to enhance the Wind Profiler performance A new Data Acquisition Card is used now to increase the dynamic range of the radar New data acquisition and data processing programs are developed in order to have an operative radar Also a graphical user interface is created to make the Wind Profiler operator tasks easy including making all the required operations and needed files in a single window easy to manage Finally new antenna shrouds are used to improve the antennas isolation Tests have been done showing improvement that led to optimistic future results A 8 2 Future work The first thing that needs to be done to the Wind Profiler is to fix the synchroniza tion issue with the Data Acquisition Card A proper synchronization will allow the computation of proper doppler information without having to substantially increase the complexity of the signal processing New antenna isolation measurements should be done with the shrouds Mounting the antennas and the shrouds in a platform to maintain alightment and pointing and making the system mobile Also adding absorber to the transmitter s shroud edge could improve the antenna isolation Then the receiver can be modified adding more gain in the front end to take advantage of the antenna isolation improvement and so reduction of the received leakage optimizing the Wind Profiler s performance After having checked the cor
24. a 3e8 915e6 74 prof avg 10000 here we decide how many profiles we re gonna average size sample 8 A double has 8 bytes so does a sample number channels 1 number of channels on the file to process not on the board sample freq 60000 sample frequency outfile infile fig that should be modified later for ordinary data ch 0 By now only one channel prf 100 on ioerror at eof openr lun noise gt Users user Desktop noise floor IDL Default noise 5min ampnevfilter2 img fig print noise opened openw lun out outfile openr lun infile info fstat lun Block size double 0 readu lun Block size total samples info size size sample total profiles total samples 1 Block sizexnumber channels print Total profiles in file total profiles once it is known how many profiles there are on the file it is proceed to process it T f frame dblarr Block size prof avg rawdata dblarr number channels Block size wsamples Block size windw rebin hanning wsamples wsamples prof avg winfunc dblarr Block size prof avg winfunc 0 wsamples 1 windw First noise mean estimated from 10 minutes noise file print fstat lun noise pow1_noise dblarr 199 readu lun noise powl noise mean noise mean powi noisel50 1981 the mean is found substracting the first samples because at low frequencies there s no noise due to the filter response print mean mean
25. ant in throughout the bandwidth 5 3 Minimum Detectable Signal After achieving the improvement of the Wind Profiler sensitivity thanks to the good estimation of the noise floor it is interesting to find out which is the minimum signal that can be detected with the new noise floor level achieved To do so the cal ibration signal is attenuated until it is not detected and the penultimate attenuation before the Wind Profiler is unable to detect it will be the one used to calculate the minimum detectable signal 32 signal noise amp estimation 30 50 E 60 a 1 1 AA AAA 1 Li 1 o 50 100 150 200 number of samples Figure 5 3 Noise floor estimation results with estimation of leakage method In Figure 5 4 the signal before subtracting the noise floor can be observed In it there is apparently no calibration signal But in Figure 5 5 after subtracting the noise floor it is seen how the calibration signal is now detected Two small tones appear in the plot showing the direct and the triple path through the BAW delay line that it was desirable to detect This minimum detectable signal corresponds to a 150 dBm signal That minimum detectable signal is not good enough yet because only scatter from strong echoes such as the ones occurring in rain and snow or high C occurring in hot humid weather can be detected In any case the Wind Profiler is still being upgraded and that minimum detectable signal w
26. atafile lun value 0L widget control ev id get uvalue value case value of radar stop begin 84 print radar stop close 1 For testing purposes debug etc etc close 2 For testing purposes debug etc etc reset return end radar start begin print radar start get conf file ev create the conf file run windprofiler executes windprofiler c and inits radar fpga return end data stop begin print data acquisition stop Spawn sudo killall s TERM fmcwsa return end data start begin print data acquisition start run daq inits data acquisition return end display data begin 85 print displaying data windprof oldaq windprof oldaqdisp return end else return endcase return END D 3 get conf file pro pro get conf file ev 3 Creating the fmcw conf file openw 1 Users user Desktop PFC IDL FMCW WindProfiler Last fmcw prueba conf id widget info ev top find by uname centr freq widget control id get value centr freq 86 printf 1 Center Frequency MHz centr freq printf 2 Center Frequency MHz centr freq id widget info ev top find by uname bw widget control id get value bw printf 1 Bandvidth MHz bu printf 2 Bandwidth MHz bw id widget info ev top find by uname chirp time widget control id get value chirp time printf 1 Chirp Length ms chirp time printf
27. beat fe cor responding to a fixed range different depending on the Wind Profiler s configuration Some of the important specifications of the delay line are showed in Table 3 2 One of the most important characteristics is the triple travel suppression with a minimum of 11 dB of suppression That means that other fixed frequencies beat will appear at 3f and 5f 21 Parameter Value Center frequency 915 MHz Bandwidth 25 MHz Time Delay 2 000 0 012 s Insertion Loss at 915 MHz 11 5 dB Tripple Travel Suppression 11 dB min Peak Power 100 mW Table 3 2 Delay line specifications 3 2 3 Receiver Subsystem The receiver subsystem consists of three identical RF paths for the three receiver channels The received signal is a mix of atmospheric echoes clutter and antenna leakage That leakage is orders of magnitude larger than the echoes and it can easily saturate the front end of the receiver The calibration signal is also coupled to the received signal in the front end ET o EN Figure 3 10 Receiver subsystem box 22 The first component of the chain is a low noise amplifier in order to ensure a good noise figure Doing that the noise figure of the system will primarily be the noise figure of the first few components in the receiver chain The coupler in the Wind Profiler is a 20 dB coupler from Narda Microwave East with insertion loss of Lcoupler 0 3 dB The power level of the d
28. distorting The maximum gain achieved was approximately 85 dB but the filter was still introducing a great amount of distortion The solution was to improve the dynamic range of the receiver somehow in order to be able to relax the specifications of the audio module To achieve that a new Data Acquisition Board was bought The board is a four channel low noise 24 bit delta sigma PC104 Plus from General Standards With that new board the dynamic range was improved from 16 bits to 24 bits which means a theoretical improvement of 24 dB 25 4 1 24DSI6LN Data Acquisition Board The selected board for the new Wind Profiler configuration is the 24DSI6LN board from General Standards The board has 24 bits and 4 channels It also has low noise features due to its Delta Sigma architecture Delta Sigma architecture is based on the technique of oversampling to reduce the noise in the band of interest which also avoids the use of high precision circuits for the anti alias filter In Delta Sigma converters noise is further reduced at low frequencies which is the band where the signal of interest is and it is increased at the higher frequencies where it can be filtered This technique is known as noise shaping That feature is also a disadvantage because due to that oversampling the external input clock that can be used needs to be very fast specifically from 25 6 Mhz to 51 2 Mhz With the former Wind Profiler the old DAQ used a totally synchronized
29. e DIO xoi ate ees draco oa de ao NOD e Ux 86 D TUNA Pro Er pro au deu eere rac een ee ant ee e and 88 DEUS A bo da a iate dude ont 89 DO TIKANG DIOS sc d A E E ee Bek IE GE OE ES ONS Y TI 89 BIBLIOGRAPHY meto bar px b b y Se D e amoa a M cao BA 91 vil LIST OF TABLES Table Page 3 1 Initial System specifications ee 0 BI n 10 3 2 Delay line specifications ax en peso o cule ne Eee son xe ra wt 22 viii LIST OF FIGURES Figure Page 21 Boundary layer structure from 6 42o aaa det eta 5 2 2 Time versus frequency diagram a FM signal b 7 2 3 Difference on instantaneous frequencies on FMCW 8 3 1 Wind Profiler old configuration ect ewe s awk eee ee RD RR 12 3 2 audio amplifier ISSUE ola bee d o bd eda b de Be o tae 13 3 3 Shroud fence built around the Wind Profiler antennas 14 3 4 Antenna isolation and achieved transmit leakage cancellation 15 3 5 New FMCW Wind Profiler block diagram 17 3 6 Radar deployment in the laboratory napana za eR RE XY v 18 3 7 Altera Cyclone II EP2C20 FPGA ck wid oed E eod pite Ru 19 3 8 FPGA Control signals generation tae 20 3 9 Control and transmit subsystem DOX Vinicio ad 21 3 10 Receiver subsystem Docs aces ur beer a We ma bhe a 22 3 11 Audio module filter ringing us ceo hoa X poe 24 4 1 2ADSIGLN Data Acquisit
30. e and high gain to compensate for the low gain in the front end So improving the Wind Profiler s antenna isolation will boost its performance Improving the antenna isolation is complicated because the two antennas are next to each other and so far field theory cannot be applied Near fields are difficult to study and simulate The solution chosen to improve the antenna isolation is the use of new antenna shrouds created for that purpose The design of the shrouds is based on a NOAA design 11 The metallic shrouds are 5 feet tall and they have openings that are cut at 45 degrees 35 6 1 Antenna Isolation Tests With New Shrouds In December 2009 the new shrouds were ready to be tested on the Wind Profiler Measurements of antenna isolation where conducted at Tilson Farm There were several sources of reflection at Tilson farm like the van were all the Wind Profiler equipment was set a big oak tree at 30 meters from the deployment point and a trailer at 15 meters See Figure 6 1 for more details Figure 6 1 Tilson Farm deployment point The transmit antenna was set as far from the van as the antenna feed cables allowed in order to avoid reflections as much as possible The receiver was set between the transmitter and the van With that deployment an antenna measurement was done without using the shrouds The leakage received is showed in Figure 6 2 being 19 dBm After that the antenna isolation was measured using the s
31. e to target as shown in Figure 2 3 It will be also attenuated and possibly Doppler shifted in case the target was not stationary The received signal will be then mixed with a replica of the transmitted signal a sinusoidal signal Figure 2 2 Time versus frequency diagram a FM signal b for every target will be produced The frequency of which is called beat frequency and depends on the range to target and the radar parameters For stationary targets the relation is as follows cT Ho 22 2 2 where R is the range to target c is the speed of light B is the chirp bandwidth and T is the sweep time Fast Fourier transform analysis is performed then in order to convert frequency information to range The range resolution of the system is Ake 2 3 2B l Parameters such as B and T are configurable in an FMCW radar thus providing flexibility to adapt to the most suitable mode without change in the peak transmit power for a given sensitivity Frequency of received signal x Time Figure 2 3 Difference on instantaneous frequencies on FMCVV To retrieve velocities it is possible to determine Doppler velocities comparing the changes of phase of two consecutive pulses received The maximum unambiguous velocity that can be measured is determined by the radar operating frequency and the pulse repetition frequency as Urmax 477 2 4 To maximize range resolution and sensitivity both B a
32. elay line signal which is coupled into the receiver through the coupler should be such that no extra thermal noise from the transmitted signal gets coupled into the receiver The LNA is a Phoenix PA911C with a noise figure of 0 9 dB The next component in the chain is a ceramic band pass filter BPF from Lark Engineering with a bandwidth of 25 MHz and insertion loss of 2 5 dB The Wind Profiler utilizes a high power mixer from Mini Circuits ZEM 4300MH4 with a 1 dB compression point of 9 dBm The mixer has a LO power level of 13 dBm For reliable performance the LO level should vary no more than 3 dBm from its specified value For the Wind Profiler operating frequency range the mixer insertion loss varies from 6 to 8 dB depending on the LO input level The dynamic range of the Data Acquisition Card and the expected backscattered power from the phenomena desired to be observed are the two parameters defining the total gain that the receiver should have The Data Acquisition Card in the new Wind Profiler configuration is a 24 bit 24DSI6LN from General Standards The specified voltage range is 10 V corresponding to a maximum input power of 30 dBm for a single tone The dynamic range of the DAQ is 110 dB setting its quantization noise floor at 80 dBm The audio filter in the receiver chain is a bandpass active filter designed by David Garrido Its design is totally customized for the Wind Profiler To design the filter it was important to e
33. er 915 MHz Span 50 MHz Center 915 MHz Span 50 MHz Figure 3 4 Antenna isolation and achieved transmit leakage cancellation Later on it was discovered that the AO card has noise of approximately the two LSB which was being injected to our Vector modulator Therefore improving the de sign in order to avoid using the vector modulator will improve the noise characteristics of our receiver 15 Proposed Solution The low transmission power and the low antenna gain require a receiver with low noise and very high gain To achieve that the receiver will be redesigned avoiding the use of the active cancellation loop A new audio module will be created with low noise components such as the Linear Technology Ultralow Noise Low Distortion Audio Op Amp LT1115 8 The design will be totally customized for the FMCW Wind Profiler s needs The front end of the receiver will be customized too Moreover a new Data Acquisition Board of 24 bits instead of 16 bits is required and a new modification of the FPGA design is needed In Figure 3 5 the new FMCW Wind profiler block diagram is showed Parts in color are modified from the previous design 16 ZHW Sc SSS eee E ed ZHN SZ ZHN S16 9 Figure 3 5 New FMCW Wind Profiler block diagram 17 3 13 FMCW Radar Current State The FMCW Wind Profiler radar is still in a developing stage A new audio module is currently working with the radar the front end
34. filer operates in the 900MHz ISM frequency band 902 928 MHz with a center frequency of 915MHz The chirp is generated by Direct Digital Synthesizer AD9858 with a bandwidth up to 25MHz and linear FM modulation So achieving a maximum range resolution of 6m Parameter Value Transmitter Center frequency 915MHz Peak Transmit power 30W Transmitter type Solid State RF SSRF Sweep bandwidth lt 25MHz Sweep time 8 333ms PRE 100Hz Antennas Type Four Parabolic dish Gain 18dB Polarization Linear Front to Back Ratio 22dB Table 3 1 Initial System specifications The default mode of operation had a PRF of 100Hz and a duty cycle of 83 3 T 8 33ms These parameters allow detection of vertical unambiguous velocity up to 8 25m s which exceeds the range of usual turbulent velocities on the boundary layer 3 5m s All the signals needed to start the radar such as those used to start and set up the DDS and synchronize and clock the sampling are generated by an FPGA Cyclone II from Altera designed by Iva Kostadinova and David Garrido Lopez The parameters of the radar such as bandwidth PRF and T are configurable This allows for adaptation to the resolution and scenario as necessary When 2 3 2 4 and 2 5 are taken into account it is possible to calculate the optimal configuration The transmit amplifier is a compact solid state RF power amplifier prov
35. he relationship between the radar volume reflectivity and refractive index structure as mA 0 382473 2 1 For a given radar A the radar reflectivity n of a region of refractive index fluctu ations is directly proportional to C when the length scale of one half of the radar wavelength falls within the inertial subrange The more violent the turbulent mixing the larger the displacements and the stronger the inhomogeneities will be So strong turbulence and sharp mean gradients contribute to high C values The radar backscatter can be originated by other sources such as Rayleigh scat tering for example from birds and insects the size of which should be much smaller than the radar wavelength and is proportional to A7 In our application this kind of scattering will be considered as undesired noise 2 3 FMCW Radar Basics FMCW is common technique used on radars that avoids the limitations of pulsed radars It consists on the transmission of a sinusoid whose frequency changes over time in our case linearly Ideally the instantaneous frequency should augment indefinitely with time but in order to have a realistic system the frequency will increase until a maximum value and will start from the initial frequency once reached that point So the instantaneous frequency will have a saw tooth shape as observed in Figure 2 2 The backscattered signal will be delayed by taetay AR where c is the speed of light and R is the rang
36. hrouds Between the edges of the antennas shrouds there was 0 61 meters of separation and between the van and the receiver antenna shroud edge there was a separation of 3 meters The openings of the antennas were not parallel to each other to reduce the cross talk between the antennas The deployment can be observed in Figure 6 3 The results 36 Meas Tools O O Next Peak Next Pk Right Figure 6 2 Received leakage without using shrouds of the test were successful having 34 dBm of leakage in the receiver s front end see Figure 6 4 That means an improvement on isolation of 15 dB The obtained results are optimistic 15 dB of improvement is a great result having into account that this measurement was still experimental The next step will be to obtain the maximum antenna isolation improvement that can be achieved To do so several improvements can be made The deployment point can be changed to a place with fewer sources of reflection Absorbers can be used in the edge of the transmitter antenna shroud avoiding refraction and diffraction at the shroud edge and so improving the isolation Also the radar antennas need to be mounted on a platform to maintain alightment and pointing The shrouds openings should also be covered with plastic since these will enable the snow and water to slide off and not attenuate the signals The improvement achieved will allow more gain in the receiver s front end and in the audio
37. iding a 30W output The antennas are 4 diameter antenna parabolas with dipole antenna feeds 19 degree beamwidth and 18dB of gain The antennas have a broad beam in order to employ a spaced antenna technique for estimating horizontal winds That broad antenna beamwidth has its problems The main problem associated with it is inadequate isolation between transmitter and receiver The transmitter leakage 10 received is powerful enough to saturate either the front end of the receiver or the Data Acquisition Board To solve that problem an active cancellation loop was introduced using a vector modulator AD8340 a replica of the leaked signal with opposite phase is coupled to the receiver in order to cancel the leakage 11 idi x A 8 02 BgP Z 9 VNI 8 2 0 VN HWSCsM8 ZHINS 1625 348 SP0r 9 dwy apny ZHINSZ MG 2 4516 2 448 ZHINS 165 448 La r XC epoz PEZ O VN lnpoyy uonejjeoue 8peL 9 PuUUSJUY usueJ 80 1 8 0 dwy olpmy z Al Sg J gpoz 4 Beah popz9 dwy apay ZHNSZ M8 ZHNA9Z E 0 8P09 9 VdSS DO uz ova tamod 9 91 ZHA9Z E 0 ddd apmy 7HY9Z 0 3d8oipny 3d9 opny r inpow ojpny V9dd ZHWOL s03 A eubig gt ILLUS ZHINOOZ 04 Joyesquag jeu ig zx gt ZX saa ZHIASZ M8 402 8062 YNI 2 4916 2 388 KK Figure
38. ied at any time making the radar flexible to operate in different contexts USB Me Lime Lie WGA Video 8 232 Tan iras Oa Pon Sarai Pot A 1 ma GEI NM T m Z A Po P VU Wc wu S gt 4 b PBDasaa E Hester 2 12 Y nm with Part Prom LE s AA m e Espace Faster t UPI lu2 25 A a USD Master C oth Lhpasi hein io ba ot pe at on Dem BIN Ares a boron Cyrene M FPGA nah M6 LEI 10 Fas LEDs BEE HE 3 Green LEDs SMA ena Cheb 10 Toggle Zwiiches 4 Pust butter votiches G y SORAN GIDOR TRA Byte Flach Memory Figure 3 7 Altera Cyclone 11 EP2C20 FPGA The FPGA design was developed by Iva Kostadinova and David Garrido The design is characterized by generating all signals in total synchronization giving no room for timing skews that can lead to phase wrapping in the received data A later modification of the FPGA was developed by David Garrido in which the duty cycle of the Wind Profiler was enhanced to 10096 and the signals generated were modified 19 to adjust to the last Wind Profiler configuration For more information about the FPGA modification design refer to 2 The transmitted chirp is generated by a 10 bit Digital Signal Synthesizer from Analog Devices AD9858 DDS The DDS receives the needed parameters for fre quency sweep generation see Figure 3 8 from the control subsystem The board uses an external 800 Mhz clock referenced to the 10 Mhz FPGA c
39. ill improve soon after the antenna isolation is improved and with that improvement the receiver will be optimized to adjust to that antenna isolation improvement increasing gain in the front end between other possible modifications to enhance its performance 33 10 Powar dBm o 5o 100 150 200 samples m Figure 5 4 Minimum detectable signal plot Powar dBm o 50 100 150 200 samples E Figure 5 5 Minimum detectable signal after subtraction of the noise floor 34 CHAPTER 6 ANTENNA ISOLATION One of the main problems with the FMCW radars is related to the isolation between the antennas In FMCW radars transmission and reception is done at the same time using different antennas for transmission and reception unlike pulsed radars that transmit and receive at different times through the same antenna Pulsed radars have height limitation at low altitudes due to their pulsed architecture FMCVV radars have very high resolution and small peak to average power but the isolation between the antennas is an important issue In the Wind Profiler case the isolation between the antennas is 71 dB but a great amount of leakage is still coupled to the receiver thorough the side lobes That leakage is a limiting factor for the Wind Profiler s performance not allowing the addition of enough gain in the front end of the receiver to avoid saturating the mixer and requiring a complex audio module with high order to filter the leakag
40. ion Card ca EIC ERA 26 5 1 5 2 9 3 5 4 5 5 6 1 6 2 6 3 6 4 7 1 Leakage and calibration signal processed 31 Noise floor estimation results ici Kk 32 Noise floor estimation results with estimation of leakage method 33 Minimum detectable signal plot ccs oe sa yaa ke to ka eo ap o te Yad 34 Minimum detectable signal after subtraction of the noise floor 34 Tilson Farm deployment Poltica aaa 36 Received leakage without using shrouds 37 Wind Profiler deployment with shrouds 38 Received leakage using shrouds 6 200280 dl see Uu xe hs 38 Graphical Usera oo da b b b s 40 CHAPTER 1 INTRODUCTION 1 1 Motivation The measurement of winds and processes taking place in the atmosphere is a fun damental requirement in both research and operational meteorology This project is focused on the processes taking place in the lower troposphere called the atmospheric boundary layer ABL The ABL is important meteorologically in terms of assessing of convective instability The entrainment zone at the top of the ABL acts as a lid on rising and cooling air parcels due to temperature inversion An external mechanism such as geographically forced uplift vigorous surface heating or dry lines can break the entrainment layer allowing the capped air parcels to rise freely As a result vigorous convection
41. is as follows At the beginning of the program the different parameters needed for processing such as frequency sample number of profile aver age and name of the files are configured Then the synchronized data from the input file is stored in arrays A Fast Fourier Transform and the chosen profile average are done In the next step range and doppler velocity information are extracted and written to a file Figure 5 1 shows a processed signal containing leakage and calibration signal It can be observed that there is no trace of leakage whatsoever and that three different tones appear The tones are due to the architecture of the BAW delay line The first tone correspond to the direct path of the delay line input signal through the delay line the other two tones correspond to triple and the quintuple path through the delay line The plot also shows that the noise floor is the limit for the Wind 30 Profiler sensitivity To improve the sensitivity a estimation of the noise floor can be subtracted from the processed data 20 Powar dBm o 5o 100 150 200 samples m Figure 5 1 Leakage and calibration signal processed 5 2 Noise Floor Estimation In order to improve the sensitivity of the Wind Profiler an estimation of the noise floor can be subtracted from the data The longer the profile average is the greater the improvement in sensitivity will be specifically the theoretical improvement is related to the following exp
42. le for the board to access Returned 56 gt 0 The number of errors encounterred R K SR k k k k k k k ok kk k ok k k k k k k k OA k k k k k k k k k k k k k k k static int read data int fd int errs long get sizeof buffer CHANS_ENABLED int got gsc_label Reading got dsi dsl read fd buffer get if got O 1 1 else if got get 1 errs 1 printf FAIL lt got ld samples requested 1d n long got long get else errs 0 57 printf PASS 714 samples Wn long get return errs SK k k k kkk k kk k k kk k k k k k k k k k k k k k k k k k aK ok k lok Function save data Purpose Save the read data to a text file Arguments fd The handle for the board to access chan The number of channels errs have there been any errors so far Returned gt 0 The number of errors encounterred 58 R k SR k k ok k k k ok k k k k k k k k ok k k k k k k k k ok k k k k k k k k static int save data int fd int chan int errs t FILE file int i long 1 const char name data txt long samples sizeof buffer CHANS ENABLED gsc_label Saving for if errs printf SKIPPED errors n errs
43. lock The analog mixer mixes the clock signal with the chirp centered at 115 Mhz to produce the de sired product at 915 Mhz at the DDS output The undesired mixing products are filtered then and the signal is amplified A 4 way splitter is used to split the trans mitted signal to the three LO inputs of the receiver mixers and to the RF high power amplifier The high power amplifier is a linear power solid state RF amplifier from OPHIRrf with a gain of 50 dB Finally a 20 dB directional coupler delivers the output of the amplifier to the antenna and couples part of it to the calibration loop DDS Control Signals Center Frequency Bandwidth PRF 10096 duty cycle Reset Wind Profiler Configuration File Center Frequency 115MHz Bandwidth 25MHz PRF Hz 10 FPGA DAQ Control Signals 1 Trigger Synchronization Figure 3 8 FPGA Control signals generation 20 Figure 3 9 Control and transmit subsystem box 3 2 2 Calibration Loop The calibration loop allows continuous monitoring of the system performance and gives a reference point with which to evaluate the radar sensitivity The delay line used is a bulk acoustic wave BAW delay line with an effective delay of 2us from Teledyne Electronic Devices The input of the delay line is an attenuated replica of the transmitted signal The output of it is coupled into the receiver chain through a 20 dB coupler The delayed chirp produces a signal of a fixed frequency
44. module translating to a better radar minimum detectable signal 3T Figure 6 3 Wind Profiler deployment with shrouds Ref dBm Mkr1 925 98 MHz Peak Atten 10 de 34 81 dBm Log 18 Clear Writ Marker y Max Hold 925900009 MHz 3481 dBm Figure 6 4 Received leakage using shrouds 38 CHAPTER 7 GRAPHICAL USER INTERFACE The operation of the Wind Profiler requires knowing how to start configure and manage all the different parts of the radar For example start the Wind Profiler sending to the FPGA the required parameters start and stop the data collection and process and display the acquired data In order to make the Wind Profiler operation easier for future users a graphical user interface GUI was created using IDL The purpose of it was to manage all the Wind Profiler operations commented on above using a single window with a friendly and easy to understand interface During the Spring of 2009 the Wind Profiler GUI was developed for the old configuration To adapt the GUI to the new Wind Profiler configuration only a few changes such as path and filename modifications will be required when all the programs and files required for the new configuration are definitive The GUI interface is presented in Figure 7 1 The GUI allows to modify the Wind Profiler parameters directly from the interface instead of having know where the conf file is open it and modify it manually When Radar Start button is p
45. nd T should be as large as possible The value of T however is constrained by the coherence time of the atmo spheric echo because the presented theory is based on the assumption that the target produces constant frequency sinusoidal echo during the sweep 12 The maximum range for FMCW radars is determined by the sweep time T and the sampling fre quency used in the A D conversion The latter gives us the maximum beat frequency that can be detected without aliasing CHAPTER 3 SYSTEM HARDWARE DESCRIPTION 3 1 System Overview The current FMCW radar project started as an analog of the S Band FMCW boundary layer profiler developed in 2003 at University of Massachusetts at Amherst 12 The change from S Band to UHF was proposed in order to reduce the Rayleigh scattering from insects and birds which appeared to dominate the observed vertical profile of the mean reflectivity at S Band The current FMCW Wind Profiler was started in the summer 2006 by the former students Albert Genis during 2008 and Iva Kostadinova on 2008 2009 Many changes have been introduced to the radar after upgrading and overcoming the different prob lems that were affecting the radar through several years The purpose of the next sections is to explain the latter hardware configuration of the radar and analyze the problems and changes that either were introduced or are being introduced to solve them 3 1 1 Initial hardware configuration The FMCW wind pro
46. nsure that the audio filter significantly attenuates the leakage signal and there is a trade off between filter ringing and desired observable minimum heights Because one of the primary design goals of the Wind Profiler is to be able to 23 observe boundary layer processes at low altitudes it is important to make the trade off carefully In Figure 3 11 the filter response to a signal composed of leakage plus calibration signal can be observed Figure 3 11 Audio module filter ringing 24 CHAPTER 4 DATA ACQUISITION After identifying the problems encountered with the Wind Profiler it was nec essary to proceed to modify the receiver and to create a new audio module totally customized for the Wind Profiler that will enhance its performance The design was done employing low noise components to achieve a better performance for example the operational amplifier used was the LT1115 from Linear Technology 8 with a typical voltage noise of 0 9nV Hz at 1 kHz With the 16 bit Power DAQ PD2 MFS 4 500 16 from United Electronics Indus tries Inc that the former Wind Profiler configuration used the gain needed for that audio module must be at least of 110 dB and the order required was order 10 During the Summer of 2009 some audio module prototypes were built That work showed that it was not possible to meet the specifications needed for the Wind Profiler The high gain and the high order of the audio module made it unstable and
47. ofiler int fd 1 int errs gsc_label Input Channels errs dsi query fd DSI QUERY CHANNEL QTY chans if errs 0 printf ld Channels n long chans 0 return errs 68 APPENDIX B SYNCHRONIZATION PROGRAM B 1 pre processing david pro Pre processing Consist on synchronize the data and output a file with only data and a mark in every profile ih channel contains info and channel 3 contains sync ih Save info with a mark every edge of sync signal and save it in a new file dataMark txt Developed by David Garrido pro pre processing david infile ch sync number channels 6 currently the board has 6 channels fs 60000 define the frequency sample PRF 100 outfile infile img that should be modified later for ordinary data print outfile lun 1 Our input file lun_out 2 our output file 69 close all on ioerror at eof jump to at eof in case of ioerror 3 Read the adquired data from the infile organized in columns Default route Users user Desktop noise floor IDL Default test dat print before data lines skipped 0 define the number of lines skipped at the beggining of the file the structure of the data is defined choose between the available data readcol infile F FMT chO ch3 SKIPLINE lines skipped the columns are saved we can discard data using X in the format instead of A chO strmid ch0
48. rect performance of the Wind Profiler real data should be acquired from field deployments After processing the data collected the performance of the radar can be checked and modifications can be added in case the data collected is not satisfactory Finally when the Wind Profiler is ready slight modifications to the graphical user interface and to the real time display programs can be performed to adapt them to the new Wind Profiler configuration In a second phase it is planned to use a space antenna technique 9 to be able to analyze horizontal winds too These techniques can obtain more rapid wind estimates compared to Doppler beam swinging systems By making all of the changes mentioned previously the Wind Profiler should be able to operate properly 42 APPENDIX A DATA ACQUISITION PROGRAMS A 1 main c Rev 1814 Date 2009 04 22 17 13 15 0500 Wed 22 Apr 2009 Program developed by David Garrido Lopez include lt ctype h gt include lt stdio h gt include lt stdlib h gt include lt string h gt include lt time h gt include main h Hdefine TRIGGER_RATE 100 Hdefine SAMPLE_RATE 60000L define CHANS_ENABLED 6 variables eb RRO k k E K KK k kK K K kk kk static int _continuous 0 static int ignore errors 0 43 static int index 0 static long minute limit 0 static __s32 buffer data 3600L E kkk k kkk k kk k kkk k kkk k KK k KKK k k k k k ok k
49. resented here and conclusions and recommendations for future work are drawn CHAPTER 2 FMCW WIND PROFILER PRINCIPLES 2 1 Atmospheric Boundary Layer The boundary layer is the lowest 1 2 km of the atmosphere the region most directly influenced by the exchange of momentum heat and water vapor at the earth s surface Turbulent motions on time scales of an hour or less dominate the flow in this region transporting atmospheric properties both horizontally and vertically through its depth The mean properties of the flow in this layer the wind speed temperature and humidity experience their sharpest gradients in the first 50 100 m appropriately called the surface layer Turbulent exchange in this shallow layer controls the exchange of heat mass and momentum at the surface and thereby the state of the whole boundary layer It is hardly surprising we should have a lively curiosity about this region There are two main types of boundary layer The convective boundary layer where heat from the Earth s surface creates positive buoyancy flux and instabilities that lead to turbulences and the stably stratified nocturnal boundary layer where negative buoyancy flux decreases the turbulences and stable stratified conditions prevail The ABL can reach over 3 km during daytime while the usual height at night is between 50 m to 300 m Typical boundary layer structure is depicted in Figure 2 1 Free atmosphere K H waves
50. ressed the parameters introduced in the left part of the GUI are stored and the Wind Profiler is started using these parameters It is started by executing the c program that processes these parameters and sends all the necessary information to the FPGA through the serial port When Radar Stop button is pressed the FPGA is reset and with that the Wind Profiler stops The buttons Start Data Collection and Stop 39 000 X UMass FMCW Radar GUI v1 0 iE E paa pa zm 2 Eu Lm m l Bl Figure 7 1 Graphical User Interface Data Collection execute and kill the data acquisition program respectively Finally the Display Data button executes the data processing program with the specified time average and concurrently executes the real time display program developed by David Garrido to display the data while its being collected and processed The GUI files are provided and commented in Appendix D 40 CHAPTER 8 CONCLUSIONS 8 1 Summary This thesis has described the FMCW Wind Profiler state as of Fall 2009 The previous configurations and problems have been explained and then solutions pro vided Thanks to the previous work done with the radar the sources of error can be identified and the radar receiver was upgraded in order to improve its performance The receiver s front end was modified and the leakage cancellation loop used before is now avoided Also a new audio module was specifically designed The F
51. ression 10log Nprofites In a first attempt it was proceed to estimate the noise floor in a traditional way capturing the output signal of the Wind Profiler when no input signal is being introduced to the receiver The receiver inputs were terminated at 50 Ohms the calibration loop input and the LO port of the mixer were also terminated at 50 Ohms The noise floor estimation obtained was not good enough At low frequencies the big differences between the real noise floor and its estimation make sensitivity improvement not possible In Figure 5 2 the first noise floor estimation and its result is showed Blue line is the received signal red line is the noise floor estimation and black line is the result when the estimated noise floor is subtracted to the signal 3l The results obtained were not as good as it was expected the audio module output T T T T T T T Powar dBm 100 samples m Figure 5 2 Noise floor estimation results noise floor is different depending on its input signal Because of that it is necessary to make an estimation of the leakage and introduce it in the receiver s input in order to obtain a good estimation of the noise floor The commented method turned out to be a valid procedure to estimate the noise floor In Figure 5 3 the results obtained with that method are shown Now in this case the improvement of sensitivity is achieved in all the bandwidth of interest and that improvement is also const
52. rs gsc_label_level_dec return errs k k kk OO k k k k K k ok k k k k k k k k K K ok Function dsi config board profiler Purpose 62 Configure the given board using common FMCW defaults Arguments fd The handle to use to access the driver index The index of the board to access Ignore if O fref This is the PLL Fref value or 1 to use the default fsamp This is the desired Fsamp rate of 1 to use the default Returned ko gt 0 The number of errors encountered here R R SR k k k ok k k k k ok k k k k k k k k k k k k k k k k k k k k k k k k k kkk s32 fref 3 int dsi config board profiler int fd int index 1 s32 fsamp int errs 0 832 ndiv 832 nrate Legacy 832 nref PLL 832 nvco PLL errs dsi rx io mode fd index GSC IO MODE DMA NULL 63 errs errs errs errs errs errs errs errs errs errs errs errs errs errs errs errs errs dsi rx io overflow fd index DSI IO OVERFLOW CHECK NULL dsi rx io timeout fd index 30 NULL dsi rx io underflow fd index DSI IO UNDERFLOW CHECK NULL dsi initialize fd index dsi ain mode fd index DSI AIN MODE DIFF NULL dsi ain range fd index DSI AIN RANGE 10V NULL dsi ain buf input fd index DSI AIN BUF INPUT ENABL
53. s the radar spawn gtt o Users user Desktop PFC IDL FMCW WindProfiler Last windprofiler Users user Desktop PFC IDL FMCW WindProfiler Last vindprofiler c print windprofiler compiled spawn sudo Users user Desktop PFC IDL FMCW 88 WindProfiler Last vindprofiler fmcw conf print windprofiler executed end D 5 reset pro pro reset 3 Compile and execute reset c which stops radar spawn gtt o Users user Desktop PFC IDL FMCW WindProfiler Last reset Users user Desktop PFC IDL FMCW WindProfiler_Last reset c print reset compiled spawn sudo Users user Desktop PFC IDL FMCW WindProfiler Last reset print reset executed end D 6 run daq pro pro run Daq 3 Compile and execute fmcwsa c that inits the data acquisition 89 spawn gtt o Users user Desktop PFC IDL FMCW RunDaq fmcwsa Users user Desktop PFC IDL FMCW RunDaq fmcwsa c print fmcwsa compiled spawn sudo Users user Desktop PFC IDL FMCW RunDaq fmcwsa fmcw conf print fmcwsa executed end 90 BIBLIOGRAPHY 1 David Atlas Radar in Metrology American Meteorological Society Boston 2 10 11 12 l 1990 David Garrido Lopez An UHF Frequency Modulated Continous Wave Wind Profiler Receiver and Audio Module Development Master s Thesis University of Massachusetts at Amherst 2009 David M Pozar Microwave and RF Design of Wireless Sys
54. synchronization is being fixed 4 2 Data Acquisition Program The DAQ needs a C program called savedata c to collect data This program configures the board according to the desired parameters first doing a test and after that acquiring data to a file until either the timeout specified when the program started is reached an error occurred or when the program is stopped manually by pressing CTRL C in the keyboard This program uses some other files that are not included in this document main h savedata h between others The main files savedata c and main c are provided and commented in Appendix A This code also uses some specific functions for this device Further information about these functions can be found in the 24DSI6LN linux driver manual 5 The structure of the program is as follows First the board is searched identified and selected After that the device is opened some data is collected in a brief test and if the test is successful it proceeds to capture the real data The board is configured using the function dsi config board then the file which will contain the data is created using as a filename the current time After the file is created the program continuously collects the data until it is terminated Before executing the program it is important to have started the driver To start the driver login as a root user change to the directory where the driver was installed
55. tems John Wiley amp Sons 2001 Earl E Gossard Refractive index variance and its height distribution in different air masses Radio Science 12 1977 General Standards Corporation 24DSI Linux Device User Manual Datasheet Iva Kostadinova An UHF Frequency Modulated Continous Wave Wind Pro filer Developement and Initial Results Master s Thesis University of Mas sachusetts at Amherst 2009 Liam E Gumley practical IDL programming Morgan Kaufmann San Diego 2002 Linear Technology Ultralow Noise Low Distortion Audio Op Amp LT1115 Datasheet Muschinski A S J Frasier Supplement to Investigation of Turbulence and Intermittency in the Convective Boundary Layer Using a 915MHz Volume Wind Profiler Profiler Fabrication 2006 Ottersen H Atmospheric structure and radar backscattering in clear air Radio Science 4 12 1969 1179 1193 Paul E Johnston David A Carter James R Jordan Allen B White A New Snow Level Detection Radar NOAA 2009 Turker Frasier S Muschinski A and Pazmany A An s band fmcw boundary layer profiler Description and initial results Radio Science 38 4 2003 91
56. tly the need for switches that are both faster and high powered FMCW radar alleviate this problem by using separate antennas for transmit and receive which also allows use at short ranges The problem in dual antenna systems is parallax at low altitudes due to the spatially separated antenna apertures and some uncertainty in the actual sampling volume The primary objective of this thesis is to explain the previous state of the radar providing a detailed account of it Then to present the current state of the radar with results and complete explanations to report conclusions 1 2 Summary of Chapters Chapter 2 gives a introduction to the atmospheric boundary layer explains its structure and characteristics followed by clear air backscatter theory After that the principles of FMCVV radars are explained Chapter 3 presents an overview of the FMCW Wind Profiler and the initial configuration and previous results of the radar are explained and commented Chapter 4 depicts the Wind Profiler data acquisition by presenting the Data Ac quisition Card and programs Chapter 5 explains the signal processing method used in the Wind Profiler and concludes by showing the radar s minimum detectable signal In chapter 6 the new idea to improve the Wind Profiler antenna isolation is shown and the results obtained are provided In chapter 7 the Graphical User Interface is displayed Finally chapter 8 contains a brief summary of the research work p
57. very helpful Ivona Kostadinova deserves a special consideration as well because she introduced me to the radar and did the previous work with it which helped me get to the point where I am now Finally I want to thank my family for their support from a distance Thanks to them I did not feel alone even with an ocean in between us TABLE OF CONTENTS Page ACKINOW LEDGMEN ES khu eo BING at eae OR EA iv DIST OF TABLES dul sad tesa tiec Aga tt eed ex 55 viii EIST OF ERZURES EB DURER a BG er dead ix CHAPTER Il INTRODUCTION iS n S ve e aul 1 LI OLYA vasi b yan ORE GEAR TG Pe ee eh ie aaa 1 1 2 S mmary f Oba phere Nia ee BAS ata Mie ce ee aa adi te p 3 2 FMCW WIND PROFILER PRINCIPLES 4 2 1 Atmospheric Boundary Layer l ace ei ce DeC mi Efe t cd 4 22 Clear Air Backscatter Theory oor Seas ale 5 2 2 ENGW Radar pe se d o ed do ee dire oc s s db 6 3 SYSTEM HARDWARE DESCRIPTION 9 dul System Overview eein eean E ENS esu ERE TIN Na RA 9 3 1 1 dJmialharndwareconBs ratidn eeen e ss S DP eee ee 9 3 1 2 FMCW Radar Previous Results xu 13 3 1 3 FMCW Radar Current State dinastia PETE ANA se EXC 18 3 2 Wind Profiler Subsystems eset eene e e o s ns 18 3 2 1 Control and Transmit Subsystem 4 o en t ERE Rn 19 22 Calibration Woops erica ta 21 3 2 3 Receiver Subsystem cota sez cedi xoa 22
58. waits until the synchronization channel have a transition from 0 to 1 The synchronized data will be stored in an img file that will be used for processing later This way of synchronizing the data has its problems The sampling clock used to sample the data is generated inside the DAQ because it is not possible to generate an input clock for the DAQ fast enough to meet the DAQ requirements That means that the clock is not synchronized with the system The synchronization via software relies in sampling the synchronization signal But as the sample clock is not synchronized there will be a maximum error of 1 sample Because of that from now on the only information extracted from the data will be range because to obtain doppler information good synchronization or a more complex data processing is needed That will be perfected when the Wind Profiler has been tested and is in a later stage 29 CHAPTER 5 SIGNAL PROCESSING After the acquisition of data it is necessary to process it properly to extract the required information In the Wind Profiler two different kinds of information can be extracted from the data range and doppler velocity In the following pages only range information will be plotted 5 1 Signal Processing Program To perform the data processing a program in IDL called windprof process david pro was developed The program is provided and commented in Appendix C The struc ture of the program
59. will begin producing severe thunderstorms ABL research and studies help 1 develop and improve existing numerical weather prediction models ii understand the transfer of heat water vapor and momentum between the Earth and the atmosphere iii refine the analytical description of tur bulent processes and iv quantify the absorption and emission in the troposphere which is a major factor in shaping climate on Earth The effect of the troposphere on wave propagation has also been studied extensively for the purposes of improving radio communications The main reason for FMCW radar development is the need for continuous mon itoring of the winds and fields in the atmosphere improving in situ measurements Conventional radar profiler technologies are usually able to make atmospheric mea surements of the boundary layer but preclude the lower part of the ABL around 150 meters The Frequency Modulated Continuous Wave Spaced Antenna FMCVV SA Radar that is being developed in University of Massachusetts Amherst at the Microwave Sensing Laboratory MIRSL will allow measurements of the lower part of the ABL The use of FMCW radars is introduced in order to improve the limitations of pulsed radars Pulsed radars are limited by the pulse width and switching speed of the transmit receive switches because of the use of common antennas for both functions The pulsed nature of the radar dictates a higher transmitter power and consequen
60. x 1 aux indicate the logic level so we look for 0 to 1 transitions last trig 0 for index OL size array 1 do begin in that for we look for trigger in order to synchronize data and save to the file last block should be discarded to do so last trig variable is used if aux eq 0 then begin if ch sync index gt threshold then begin aux 1 if last trig gt O then begin writeu lun out ch infollast trig 110 1ast trig Block size 1 90 if index last trig lt Block size then print Sync error make sure the block are really block size long at least endif last trig index 72 endif endif else begin if ch sync index lt threshold then aux 0 endelse endfor print Marked file created at eof close lun close lun out Block size ch_sync ch_info print File created end 73 APPENDIX C SIGNAL PROCESSING PROGRAM C 1 windprof process david pro Developed by David Garrido Lopez UMass Amherst MIRSL 2009 That program process the data already saved into infile jin blocks by pre processing david pro Steps for FMCW Processing 1 read header creating a structure 2 read profile and convert data to time domain using fft 3 remove temporal mean to get rid of stationary targets such as buildings and trees 4 get power and average in time pro windprof_process_david infile lun 1 lun_out 2 lun_noise 3 close lun lun_out lun_noise lambd

Download Pdf Manuals

Related Search

Related Contents

Dell 857-10569 Computer Monitor User Manual  Samsung GT-P3110 Manual de Usuario  Samsung 4,2 Soundbar H600 Lietotāja rokasgrāmata  Manual do Usuário - Frente de Caixa Store V.5.0.0  Intellitag Developer`s Kit RFID Keyboard Wedge User Manual    WinTen - Tenmast Software  

Copyright © All rights reserved. Failed to retrieve file