Portfolio of Staff Exchange Training at ASTRON 4 (Part 1)

Transcription

1 Project Building on Advanced Lofar Technology for Innovation, Collaboration, and Sustainability (Nr BALTICS) Portfolio of Staff Exchange Training at ASTRON 4 (Part 1) May, 2018 BALTICS 1

2 CONTENTS 1. Introduction BALTICS concept background Objectives Work plan 3 2. Detailed description of training Repair and testing of ASTRON backyard HBA tiles Hardware repair Tile beamforming tests using HBA-PC Introduction to hardware maintennace at Westerbork observatory Maintenance at LOFAR core station HBA-PC Improvement of HBA-PC firmware Single element control function Single Mode Orbitron Satellite tracking mode Firmware changelog HBA-PC hardware References Appendix 1 24 May, 2018 BALTICS 2

3 1. Introduction 1.1. BALTICS concept background During two-week (1st part of a one-month) Staff Exchange Training at ASTRON (in further text - internship) two secondments from VIRAC staff (in further text - apprentices) are trained in LOFAR HBA Tile control via direct tile hardware access, tile hardware maintenance and LOFAR station maintenance at Exloo as well as they are introduced to Westerbork Synthesis Radio Telescope. During this visit the VIRAC staff is introduced to a wider group of experts at ASTRON such building on a solid network of contacts after returning to Latvia Objectives During this internship the main goal for apprentices is to learn theoretical and practical principles of LOFAR hardware, gain hands-on experience of LOFAR LBA and HBA antenna element and subrack maintenance and in-depth knowledge on HBA tile control. To reach these objectives apprentices were assigned to further tests and development of HBA tile portable controller (HBA-PC), supervised LOFAR station maintenance at Exloo as well as being tasked with reviewing 3 neglected LOFAR HBA tiles in the backyard of ASTRON. In this document not all gained knowledge and competence is described since it is the task of specific books, manuals and research papers. Rather an overview of gained competences and very specific nuances that might not be described in literature are given Work plan The work plan of internship was prepared by the assigned supervisor and adjusted on weekly basis due to weather conditions, availability of staff and/or tools etc. The initial work plan was as follows: Mon: May 14th Arrival trip to Dwingeloo Tue: May 15th Introduction, Guided tour at Astron (Offices, RF lab), Get familiar with newest version of HBA-PC Wed: May 16th Read HBA-PC and LOFAR documents, test HBA-PC functionality on dummy tile Thu: May 17th test HBA tiles at ASTRON backyard using HBA-PC Fri: May 18th test HBA tiles at ASTRON backyard using HBA-PC Tue: May 22nd prepare RF measurements for the 3 HBA tiles and list of updates for HBA-PC (If any) Wed: May 23rd Visit at Westerbork Thu: May 24th LOFAR station maintenance at Exloo Fri: May 25th Hand in the HBA-PC and other tools, travel to Latvia May, 2018 BALTICS 3

4 The detailed final work plan can be seen in the submitted internship diary. The work regarding LOFAR maintenance was carried out under direct supervision of the assigned mentors, however, all other tasks were carried out independently and by receiving guidance, instructions and assistance from supervisors on daily basis and on request. 2. Detailed description of training 2.1. Repair and testing of ASTRON backyard HBA tiles Throughout this report the following designations for ASTRON backyard tiles were used: Fig. 1. Designators used for ASTRON backyard HBA tiles Hardware repair There are three HBA tiles installed in backyard area of ASTRON which were used for initial testing during prototyping and deployment phase of LOFAR. These tiles were not used for several years and the current condition of them was unknown. One of the objectives was to test and repair the tiles which allowed to increase practical maintenance experience. Second, these tiles allowed to carry out first HBA-PC tests in actual field conditions and allowed to improve the usability of this device. May, 2018 BALTICS 4

5 Fig. 2. Testing Tile-D using HBA-PC Following repairs and tests were carried out: First tests using HBA-PC showed read-back issues with multiple elements for Tile-B which seemed to change after touching summator cables. Problematic elements tested okay with direct connection to HBA-PC (using +7V HBA-FE port). After a closer inspection of summators it was found that center pin tips of the coax cables which connects to communication s ummator were corroded. After cleaning the tips by scraping off the corrosion all elements tested okay (Functions Read All and Read one-by-one of HBA-PC were used) and tile was considered to be repaired. Tile-C tested okay at the first go. There was small amounts of water present at right side of the tile, but no harm was done, so this tile was considered operational. Only this tile had coax cables going to ground from the summators, but apprentices did not find to where these cables were routed. They were left disconnected. First tests of Tile-D showed issue with element (HBA FrontEnd) 6 which manifested itself as continuous LED blinking and showing no response to HBA-PC commands. Occasionally this element tested okay but then elements 1, 2 and/or 5 showed communication issues. By testing the element directly it seemed to be functional, but due to confusion of small bug of HBA-PC present at that time (when doing LED blinking test, all RF and LNA bits are set to zero. One of the test was to set delays to zenith position and looking for expected readback value of 188, but without RF and LNA bits set this value was 60, which caused confusion), apprentices replaced this element. When doing this, it was discovered that it is not possible to program a new May, 2018 BALTICS 5

6 address for element when it has default factory address (decimal value 127). After correcting this HBA-PC bug, the default address was successfully replaced with the needed value, yet it did not solve the instability issue. After more testing by using only communications summator and connecting power to element 6 directly from +7V port of HBA-PC and considering the lower than expected tile current draw it was decided to replace the power summator. This solved all issues with tile D. Out of interest apprentices inspected the PCB of faulty summator and did not find any obvious damage although there was corrosion visible on the metal enclosure. Apprentices measured unloaded voltages of all ports and found that section for elements 1,2,5,6 had lower voltage of around +6.08V which would explain the unstable operation of these elements. Rest of the ports measured V. Fig. 3. Inspecting the faulty power summator of Tile-D Tile beamforming tests using HBA-PC First beam steering tests with ASTRON backyard HBA tiles After the tile repair, front-ends and HBA-PC beamforming functionality was tested by manually changing azimuth and elevation of each tile beam to specific direction and measuring signal spectrum at X and Y polarization RF output connectors on HBA-PC. Spectrum measurement tests were carried out with spectrum analyzer, HBA-PC and laptop at a rather close distance (1..2m) to HBA tiles, which may explain very large May, 2018 BALTICS 6

7 level of RFI even for Y-pol channel (See section of this report for more details) Large increase in noise floor of X-pol spectra are generated by HBA-PC internal switch mode power supply noise Spectrum analyzer (R&S FSU) was set in MAX-Hold mode and each spectrum was accumulated for approximately 30 seconds It must be noted that grating lobes occur at visible space and opposite side of main beam theoretically above ~120 MHz especially at low beam elevations, which is confusing in case of many unknown positions of RFI sources Indicated Az values are actual tile beam position (with side 1..4 facing to 0 degrees). In case of ASTRON backyard tiles actual geographical North position can be obtained by subtracting 45 degrees from shown Az values Fig. 4. First beamforming tests using HBA-PC and spectrum analyzer. May, 2018 BALTICS 7

8 Az:0deg, El:90deg Az:320deg, El:0deg Az:120deg, El:0deg Fig. 5. HBA-B, X-pol. May, 2018 BALTICS 8

9 Az:0deg, El:90deg Az:320deg, El:0deg Az:120deg, El:0deg Fig. 6. HBA-B, Y-pol. Although measured spectrum is very messy, it can be noticed that lowest RFI level is obtained with beam pointing at zenith (El:90 deg) position, which is expected. Also various RFI sources like DAB signals at >200 MHz changes amplitude with changing azimuth position of the beam. For comparison, Fig. 7. shows RFI spectrum which was measured with spectrum analyzer, HBA-PC and laptop inside the backyard cabin and it can be concluded that most part of the signals are generated by close proximity of above mentioned equipement. The same measurements were repeated for tiles C and D as well and obtained results are very similar. Spectrum graphs for tiles C and D are available in document baltics_may2018_hba_test_1.pdf. May, 2018 BALTICS 9

10 Fig. 7. RFI spectrum of HBA band,but with all the equipment inside backyard cabin NOAA satellite tracking tests To test beamforming capability of the tiles, satellite tracking tests were carried out using HBA-PC. For more details on implemented satellite tracking mode see section of this report. It was decided to receive signals from NOAA weather satellites due to regular passes and optimum frequency for HBA. Frequencies of NOAA satellites are between 137 and 138 MHz. Due to position of backyard tiles being +45 deg offset in azimuth relative to geographical north, it was necessary to set -45 deg (+315 deg) azimuth offset under Tile Control -> Set Az/El. Firsts tests were carried out with spectrum analyzer, which showed signal to noise ratio of up to ~50 db for NOAA satellite pass at peak elevation of ~80 deg even with receiver limited system noise (no noise floor increase was observed, when connecting signal cable to spectrum analyzer). To test if beam is actually formed and pointed on satellite direction intentional Az and El offsets of +/-30 to +/-40 deg were introduced and db signal power reduction was observed as expected, therefore we conclude that tiles and HBA-PC beamforming functions as intended. These tests were repeated with all three tiles and in all cases beamforming was successful. May, 2018 BALTICS 10

11 Fig. 8. HBA tile beamforming tests using HBA-PC and R&S FSU spectrum analyzer. Coax cables were routed from tiles to ASTRON backyard cabin which allowed to carry out measurements more conveniently and with less RFI from equipament. For last test apprentices also tried receiving the NOAA signals with SDR receiver, which allowed to successfully decode actual weather image centered in Netherlands. We used low-cost RTL-SDR receiver in conjunction with SDR# and WXtoImg software. Software was set up according to manual found here: Tests like these are very straight-forward and can be repeated every day - it could be used to demonstrate concepts of electronically steerable antenna array technology to students. May, 2018 BALTICS 11

12 Fig. 9. NOAA-19 wather satellite reception using Orbitron, SDR# and WXtoImg applications. Fig. 10. NOAA-19 satellite signal received with HBA tile B Y-pol. SDR reciever used: RTL-SDR with SDR#. Maximum elevation of pass: 45deg. May, 2018 BALTICS 12

13 Fig. 11. Decoded NOAA-19 weather image ( ) 2.2. Introduction to hardware maintennace at Westerbork observatory One day of the training was dedicated for visiting the Westerbork observatory, where apprentices were introduced to operations and equipment of telescope. It was valuable experience considering that similar type of VLBI system must be maintained in Irbene. Apprentices had chance to learn about telescope operator workplace and software, back-ends room, IF distribution system, H-maser, details of telescope mechanics, newly deployed APERTIF receiver and more. In addition to tour, apprentices helped to investigate poor fringe amplitude issue experienced in last VLBI experiment - in this case focus was on IF distribution system. Spectrum shape was measured at various points in signal path and compared between both polarization channels and expected values. It was concluded that IF level was nominal and fringe issue might as well could be caused by issues with H-maser experienced week before - it was not investigated further that day. To make sure that receiving part is not causing any issue it was decided to switch to second antenna which required the change of receiver and apprentices had opportunity to participate in this process. May, 2018 BALTICS 13

14 ... Fig. 12. VLBI receiver installation process Maintenance at LOFAR core station To improve LOFAR hardware maintenance skills, one of Exloo core stations (station CS26) was visited under supervision of Henri Meulman. Previously executed station test script results served as starting point for finding issues with antenna elements. Following repairs were carried throughout the day: May, 2018 BALTICS 14

15 Power supply module replacement in subrack Tipped over LBA antenna repaired. This issue was detected by inspecting spectrum of problematic element - X and Y polarization channels showed characteristic shift of the antenna resonance noise peaks. LBA frontend replaced due to absence of the signal at one channel. Issue was narrowed down to frontend by exchanging the X and Y cables at antenna end and observing that signal spectrum also changed accordingly. There was rust/corrosion visible near connector (See Fig. 14.), which shows that reason of damage probably was moisture. Other LBA element showed similar symptoms - no signal at one of the polarization channels. It was found that coaxial cable connector at antenna was damaged and caused short circuit. Both connectors were replaced to maintain the same cable length, in this case due to mechanical reasons - to not apply uneven pull stress of the crimped connectors. One HBA element showed no signal output. It was replaced and its address reprogrammed which solved the issue. A few of the HBA tiles had bird-picking holes in surface tarp which allowed moisture accumulation in the styrofoam resulting in damaged HBA-FE supporting structure (Depicted in Fig. 15). Since no sufficient replacement parts were available it was decided to fix the existing structure using a special type of tape. During the maintenance more practical experience of operating Local Control Unit (LCU) console was obtained. Usage of rspctl commands like --rcumode, --statistics, --hbadelays and more were applied during real process of maintenance. Fig. 13. LBA element tipped over. May, 2018 BALTICS 15

16 Fig. 14. Corroded and non-functinal LBA frontend Fig. 15. Damaged HBA element May, 2018 BALTICS 16

17 2.4. HBA-PC Improvement of HBA-PC firmware During the visit, appart to testing HBA-PC in the field, various additions and modifications to firmware were carried out. Main additions include new Single Mode, Orbitron Satellite Tracking functions as well as various improvements in Menu. Work was carried out on latest version of HBA-PC hardware developed by students from Hanze University of Applied Sciences during 2017/2018. All improvements were carried on firmware version v Single element control function Single Mode During the HBA maintenance, testing of only single element of the tile is required quite often. In original design, functions like Blink LEDs and Blink LEDs one-by-one as well as Read All and Read one-by-one were implemented which allows to control ether whole tile at once or in one-by-one modes user is forced to write or read all elements just to check one of them. Also, only LED control was implemented without possibility to control RF and LNA bits in element register, which is very useful for testing/measuring RF signal output of specific element independently. The new Single Mode function is available under Tile Control menu now, which allows user to scroll through elements 1 to 16, automatically read back contents of element register and independently set RF, LNA and LED bits - this way it is now possible to concentrate on debugging single element without touching rest of the tile. Summary of implemented Single Mode features: User can scroll through elements 1 to 16. It carries to back to one when address of more than 16 is requested and back to 16 when less than 1 is requested. Selected element address is preserved when exiting and entering Single Mode menu again. Each time new element is selected, contents of register is read back, decoded and showed on the screen. Selected element can be read back again any time with the press of rotary encoder knob. User can toggle RF, LNA and LED bits with buttons (1), (2) and (3) of the keypad respectively. Each time the bit is toggled, register is read-back and actual register contents decoded and showed - this way user can be sure if write operation was successful or it failed. In case of failed read-back??? are showed in all positions. When the tile power is ON, current consumption is shown in right corner. Combined with possibility to turn individual LNAs ON/OFF, it allows to easier debug issues related to element power supply or broken front-end amplifier electronics. May, 2018 BALTICS 17

18 Fig. 16. Single Mode output in case of single element issues. In this case tile power is OFF, so it is not possible to read element registers Fig. 17. Single Mode output in the case of healthy element. It can be mentioned, that in the case of all LNAs turned ON, total current consumption should be approximately 630 to 660 ma as shown in this case May, 2018 BALTICS 18

19 Orbitron Satellite tracking mode Usually, when maintaining and checking HBA tile, it is sufficient to measure and analyze output signal spectrum with tile beam pointed on zenith position - this includes analyzing of expected RFI signal footprint and expected noise floor levels. But to test the actual beamforming functionality ether during the maintenance or concept demonstration, actual moving signals sources from space are preferred over stationary terrestrial RFI sources, one of the reason being lower efficiency of beamforming capability of HBA at low elevations. Very convenient example of such sources from the space are satellites, more specifically low earth orbit weather and amateur VHF satellites at 137 MHz and 144 MHz bands. These satellites have sufficiently strong signals for single HBA tile to be useful, relatively low frequency means no confusion/issues with grating lobes and these satellites passes regularly, every day. New Satellite tracking mode Serial Mode was implemented under Tile Control -> Set Az/El menu. In this mode HBA-PC receives Az/El strings via USB Serial Port and updates HBA tile beam respecively. Beam is updated only if Elevation is larger than zero and ether Az of El changed value. Az/El values are calculated with free satellite tracking software Orbitron and are sent to serial port with plugin DDEOrbitronToSerial. It was decided to not carry actual Az/El calculations in HBA-PC firmware, one of the reasons being that there is no visualization of satellite pass available which You get automatically in case of Orbitron. Due to simple Az/El string format, Serial Mode could be extended in the future to track other objects, like Sun, as well. Detailed manual on how to setup and use the Serial Mode is available in Appendix 1. In case if download WEB links does not work anymore, Orbitron and DDEOrbitronToSerial installation files should be available in BALTICS project directory HBA-PC_orbitron_setup Firmware changelog This is the summary of firmware changes carried out during BALTICS May 2018 staff exchange: Added software Back buttons for convenience at Control Tile, Control Tile->Set Azi/Elevation, Control Tile->Read servers, Control Tile->LED's, Control HBA-FE menus Fixed HBA-FE address programming functionality - now it is possible to programm tile element which has default factory address 127. When using Tile Control-Set Azi/El menu multiple times, initial input values now are equal to those which were set last time, not zero. Defined names for menu IDs in \Arduino\libraries\HBA_menu\HBA_menu.h and replaced them in firmware (ino files) for better code readability. Added Single Mode function under Tile Control menu. User can scroll through the individual tile elements (1...16) with rotary encoder and set RF, LNA and LED bits with keypad. Server registers are read each time when new element is selected, when any bits are changed or Rotary switch is pressed. If read was unsuccessful,???? are shown. Server current register and current delay is also shown. When tile May, 2018 BALTICS 19

20 power is on, current consumption is shown. This functionality is implemented in void ctl_single_server(void) Added function byte writetoserver(byte server_address, byte reg_x, byte reg_y) which allows to update single tile X/Y register. Added Orbitrion Serial Mode under Tile Control -> Set Az/El menu. This mode allows to receive Az/El string from Orbitron satellite tracker DDEOrbitronToSerial plugin. Example of string format: AZ:226.3,EL: 'NewLine' must be enabled in plugin s menu. See: for more information how to install the plugin. Orbitron can be downloaded here: Each time the Az or El changes (when El>0), Tile beam position is updated accordingly. Previously set Az offset in Tile Control -> Set Az/El menu is also taken into account Known Bugs 1. Possible overflow in function steerantenna(byte azi, byte elev) (\Software\Firmware\teensy program\hba-tile_v3.3\steer_antenna.ino). Variable byte azi averflows at azimuth values larger than When exiting from HBA-FE menu to Main menu with Back or hardware Back button, sometimes cursor is not printed on HBA-FE line 3. in Tile Control-Set Azi/El sometimes it is not possible to set elevation to 90 degrees, only to maximum of 89 degrees 4. In Orbitron Serial Mode Az/El sometimes changes incorrectly (probably due to incorrectly termination of DDEOrbitronToSerial plugin). Solution is to restart Orbitron and plugin. Identified future improvements 1. Possibility to set both negative and positive Az offsets would be more intuitive HBA-PC hardware Modification of HBA-PC Reset button circuit In new version of HBA-PC there is dedicated Reset button available which function unstable. After inspecting schematic and firmware it was concluded that it is software implementation and microcontroller is reset when following lines in code are executed: //reboot teensy if (digitalread (resetpin) == LOW) CPU_REBOOT; In other words it is necessary to include these lines in all the functional loops and microcontroller must be executing correctly for Reset function to work correctly which defeats its purpose. It was decided to disconnect the reset button wire from Pin 37 of Teensy board and connect it to Reset solder pad. Now Reset button does its function all the time. May, 2018 BALTICS 20

21 Internally generated noise issue at X-pol RF out port During the RF signal measurement tests with HBA-PC, it was discovered that there is wideband noise present X-pol (It supplies tile power summator) RF output when tile power is turned ON. The noise is present even when tile itself is completely disconnected. It is obvious that noise is generated by internal mains to 48V switching mode converter. In fact, by inspecting schematic of latest HBA-PC hardware, there is no power supply filtering and/or Bias-T network at all - RF out connector is connected directly to power supply via simple capacitor. This may be explained with fact that first hardware prototype used actual RCU board which had filtering and Bias-T already built in and it was not implemented in HBA-PC board. Following pictures shows X-pol port signal spectrum at HBA frequency band without tile connected: Fig. 18. Signal spectrum with tile controller disconnected from mains. The same clean spectrum was measured at Y-pol eve with tile power ON, so it can be concluded, that there is no measurable noise generated by HBA modem electronics. May, 2018 BALTICS 21

22 Fig. 19. Signal spectrum of X-pol RF port with tile disconnected and Tile power OFF. It seems that this noise is leaking through relay contacts, which is expected, because relays used are not specified for high isolation RF applications. Fig. 20. Signal spectrum of X-pol RF port with tile disconnected and Tile power ON. There is wideband noise visible with increase of ~20 to 30 db throughout all HBA signal bandwidth. It is clear that this issue must be solved for HBA-PC to be fully usable for tile maintenance. Possible options include: May, 2018 BALTICS 22

23 Implementation of RF low pass filtering between existing 48 V power supply and/or point where voltage is injected in X-pol cable. Probably similar or the same solution already used in RCU boards can be adapted. Changing 48 V power supply solution completely - ether by using better (lower noise) switch mode power supply or using linear power supply solution. Drawback of last option would be the increase of size and weight of HBA-PC. 3. References 1. M.J Norden, LOFAR Station User Manual (LOFAR-ASTRON-MEM-239), I.I. Virtanen, Station Data Cookbook (LOFAR-ASTRON-MAN-064), 2012-May M. Norden, HBA Repair Guide (LOFAR-ASTRON-MEM-278), H. Meulman, LBA Service document (LOFAR-ASTRON-MAN-056), S. Kiewiet, J. Kosse, D. Zijlstra, HBA-Tile Portable Control, Hanze University of Applied Sciences, Date: E. Kooistra, HBA Control Design Description (LOFAR-ASTRON-MEM-175), May, 2018 BALTICS 23

24 4. Appendix 1 HBA-PC Orbitron serial mode setup manual BALTICS, May Download Orbitron satellite tracker from here: 2. Download DDEOrbitronToSerial plugin and follow installation steps: 3. Run Orbitron with in Administrator mode to be able to update orbital parameters (TLE - two line elements). To update TLE: a. Press Setup button b. Then press Update TLE files in TLE updater tab: c. There should be no download errors for satellite file required, for example weather.txt in case of NOAA satellites. 4. Select necessary TLE file, in HBA case it is probably weather.txc for NOAA satellites: May, 2018 BALTICS 24

25 5. Update Your location information in Location tab. For example, ASTRON location can be simply updated by using Grid locator designator JO32et (it will fill Lat/Lon automatically). Location should be visible on the map. +/ km error is not important in this case. 6. Select satellites in the list to the right. In case of NOAA satellites, at the time of writing NOAA-15, NOAA-18 and NOAA-19 are the only ones operational and transmitting at VHF. Satellite can be activated by clicking satellite on the map. VHF Frequency information for NOAA satellites (+/- 3 KHz Doppler): NOAA MHz NOAA MHz NOAA MHz 7. Connect computer to HBA-PC with USB cable. New serial port should be installed automatically and available afterwards (Probably COM7). 8. In HBA-PC select Serial Mode under Tile Control -> Set Az/El. Appropriate Azimuth offset should be set previously - in case of ASTRON backyard tiles it is -45 deg or +315 deg. Tile power must be ON. Screen like this should be visible: 9. In Orbitron, start DDEOrbitronToSerial in Rotor/Radio-->Driver Tab. May, 2018 BALTICS 25

26 10. In plugin window, Send to Serial (Choose HBA-PC COM port number), Send Newline and Split out Az and El must be selected. Plugin window must always be open during operation. Known, but rare bug: HBA-PC sometimes receives completely incorrect Az/El values (probably due to incorrect termination of DDEOrbitronToSerial plugin and/or COM port). Solution is to restart Orbitron and plugin. 11. HBA-PC will receive and show Az/El strings always, when in Serial Mode, but update HBA delays only when El>0 and Az or El changes value (Shows Updating... text). Values are rounded down to nearest integer which is more than sufficient in case of single HBA tile May, 2018 BALTICS 26