1. Review of the EISCAT Radar Hardware Gudmund Wannberg, EISCAT HQ EISCAT Radar School Kiruna, August 15-26, 2005

2. Review of the EISCAT Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution: the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers How hardware restricts experiment design

3. A generic radar system Transmitting antenna: G T Receiving antenna: A r A/DRX Power: P TX Timing & Control To computer Signal generator Radar target:  Physics Engineering (”Radar Controller”) (”Exciter”)

4. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution : the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

5. Timing microscale: Range and range resolution Range R = ct h 0  h 0 = c t tx time 0 t tx t(h 0 )= h 0 /c + t tx /2 RX A sample taken at t = 2 h 0 /c + t tx contains contributions from the range (h 0 -  h 0 /4.......h 0 +  h 0 /4), i.e. the range resolution is = c t tx /2 If we require a range inaccuracy <  r, the TX/RX timing must be accurate to  t  2  r/c Example:  r = 30 m  t  2 10 -7 = 200 ns

6. The Radar Controller The ”heart” of the EISCAT radar systems Controls all functions that require microscale (sub-microsecond) timing precision: –TX waveform generation and amplification –Antenna selection (at ESR) –TX/RX switching –Noise injection (for calibration) –RX frequency hopping –RX sample stream gating –various RX housekeeping functions –data dumping to crate computer Basically a big, wide RAM bank, clocked at 10 GPS-MHz (100 ns) One or more RAM bits assigned to every critical function

7. Block Diagram of EISCAT Radar Controller

8. EISCAT Radar Controller

9. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution : the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

10. EROS and macroscale timing EROS is a tcl/tk software application running on the main process computer at each site. In addition to providing the run-time user interface, EROS handles all ”slow” (> seconds) timing in the system: –loads and configures microscale-timing HW, –schedules, starts, stops and changes experiments, –controls antenna pointing, –controls the polarisers and delays the R/C master clocks as required (only at the remote UHF sites) More about EROS to follow from the designer...

11. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution: the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

12. EISCAT UHF Spectrum T K S 920 925 930 MHz Klystron passband Protected ion line spectrum Protected plasma line spectrum Unprotected spectrum (GSM) 929.0 930.5 MHz Common protected spectrum

13. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution : the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

14. f out = 4 f 1 + f 2 +20 Block Diagram of EISCAT UHF Exciter

15. Block diagram of ESR Exciter

16. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution: the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

17. How a klystron works U in U out veve + U(t) - C Beam ctrl voltage U Cin Electrons are emitted from the cathode and form a space-charge cloud around it. When the beam control electrode (”mod-anode”) is raised to a high positive DC voltage U m, the electrons are accelerated. They pass through the mod-anode and are focussed into a beam, which is further accelerated by a voltage U. The beam goes through a series of resonant RF cavities to a collector. The beam current does not reach its full value instantaneously; the parasitic inductance of the the supply leads L p introduces a time constant, which in the UHF is  10  s. A RF signal applied to the first cavity sets up a RF voltage U Cin across the first cavity aperture, which density modulates the electron beam, creating ”bunches” of charge at the RF rate; i.e. the electron beam current is RF modulated. Propagating down the drift tube, the beam excites each cavity in turn and the induced fields amplify the density modulation. At the output cavity, the beam is almost fully bunched. This cavity is connected to a load (the antenna) and the beam now gives up that part of its kinetic energy stored in the RF structure (over half of the total) in the form of a strong RF current that drives the load. The power gain (input-output) can be over 50 dB. Leaving the output cavity, a little less than half the initial kinetic energy is left in the beam and this is dumped into the collector as waste heat, which must be removed by water cooling. When not transmitting, the electron beam is shut down by applying a negative (ie repulsive) voltage (a few kV) to the mod-anode. The mod-anode voltage is generated by a modulator; we can control the modulator by the BEAMON and BEAMOFF commands. Because of the time delay between the mod-anode voltage and the beam current, we cannot start to transmit RF immediately after a BEAMON but must first wait 3-4 time constants (30-40  s) for the beam current to stabilise. collector cathode mod anode modulator R/C LpLp

19. Klystron phase pushing (I) L U in U out veve + U(t) - C The phase length  of the klystron is a function of electron speed v e :  = 2  (Lf o ) / v e where L, the drift tube length, is ~ 1.4 m (UHF) But v e is a time-varying function of the accelerating voltage U(t): v e (U(t)) = c {1 - 1/[ 1+ eU(t)/( m e c 2 ) ] 2 } ½, where m e c 2 = 511 keV  = 2  (Lf o /c) {1 - 1/[ 1+ eU(t)/( m e c 2 ) ] 2 } -½ (in radians) Beam ctrl

20. Klystron phase pushing (II) At the start of a pulse, a typical value for U B (0) is ~ 85 kV (C fully charged)  (85 kV) = 2  (1.4 * 928 exp+6 / 3 exp+8)*{1 – 1/[1+(85/511)] 2 } -1/2 = = 2  * 8.4144 or 3029 degrees (v e = 0.5147 c) As the pulse progresses, C discharges and U(t) drops. C is typically made so large that the droop is small and approximately linear over the length of a typical pulse. In the UHF, C = 32  F and  U B /  t ~ –1.3 kV / millisecond After 1 millisecond, U(1) is down to 83.7 kV and the beam now runs slower:  (83.7 kV) = 2  (1.4 * 928 exp+6 / 3 exp+8)*{1 – 1/[1+(83.7/511)] 2 } 1/2 = = 2  * 8.46586 or 3047 degrees (v e = 0.5115 c) So in this simplistic model,  is ~ 17 o longer at the end of the pulse than at the beginning. Since there are several electrodes in the tube that counteract each other, the actual difference is less than half of that...

21. Phase Pushing and Spurious Doppler Phase pushing as function of beam voltage =   /  U B Beam voltage variation with time =  U B /  t Phase retardation per unit time   /  t = (   /  U B )  (  U B /  t) [   /  t] = s -1 = Hz, i.e. the phase pushing introduces a frequency shift in the output signal. This must be corrected for – or it will show up in the analysed data as a spurious Doppler velocity, v s ! T-UHF (928 MHz):  U B /  t = - 1.3 kV /ms (measured)   /  U B = 5 o /kV (measured at TTE/Paris)   /  t = - 6.5 10 3 o /s = -18.3 Hz v s = - 2.96 m/s !!!

22. Notes on Phase Pushing /Spurious Doppler Present to some degree in all pulsed klystron amplifiers, Always causes the output signal to shift downwards in frequency, EISCAT UHF / VHF transmitters fitted with feed-forward correction to reduce the effect (but not totally remove it...), ESR transmitter has no feed-forward, but the envelope of the transmitted signal is measured through the receiver and recorded, AT ESR, USE THE ENVELOPE DATA TO CORRECT FOR SPURIOUS DOPPLER IN THE ANALYSIS PHASE ! If the envelope data shows a positive spurious Doppler shift, or a frequency shift greater than some tens of Hz (+ or -), suspect a receiver and/or exciter malfunction (for instance a local oscillator running out of phase lock) – if you cannot make the Doppler look reasonable by restarting your experiment, call for help !!!

23. Basic ESR power module raises the power level from 10 mW to 250 kW Combining four modules yields 1 MW, allows pulse-by-pulse switching between antennas

24. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution: the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

25. Antennas and radiation patterns The EISCAT UHF and ESR use parabolic Cassegrain reflector antennas. To understand how they work, we recall that the shape of the far-field radiation pattern of a uniformly illuminated circular reflector of diameter D M, operating at wavelength (the main reflector of a Cassegrain antenna), is the same as that resulting from Fraunhofer diffraction of a plane wave illuminating a circular aperture set into an infinite baffle (Babinet’s principle). The intensity of the diffracted field from the main reflector at an angle  is S(  ): S(  ) = S(0) [ D M / 2 sin  ] 2 J 1 (D M sin  / ) 2 where  is the angle between the direction of observation and the optical axis, S(0) is the on-axis intensity and J 1 is the first order Bessel function. The Cassegrain optics of the EISCAT antennas also contains secondary reflectors. These are used to illuminate the main reflectors, but at the same time they also block part of the main reflector apertures.

26. Antennas and radiation patterns The subreflector blockage can be modelled as diffraction from a circular obstacle of diameter = D s (the subreflector diameter). The composite diffraction pattern of the Cassegrain system then becomes S c (  ): S c (  ) = S(0) [ /  sin  ] 2 [D M 2 – D S 2 ] -2 {[D M J 1 (D M  sin  / )] 2 – [D S J 1 (D S  sin  / )] 2 } Example: EISCAT 32-meter UHF D M = 32.0 m, D S = 4.58 m, = 0.33 m =>  -3dB (theoretical) = 0.6 degrees When D M >> D S, the full –3 dB opening angle (FWHM) of the main lobe of the diffraction pattern is  -3dB  0.89 /D M (radians)

27. The real antenna pattern differs from the theoretical S c for several reasons: - the apertures are not uniformly illuminated (physically impossible!), - the illumination does not taper off to zero at the reflector edges, - the subreflector tripod struts shade the main reflector... Radiation patterns and transverse resolution Measured pattern of EISCAT Tromsø UHF antenna  -3dB (actual) = 0.7 degrees The main beam opening angle determines the transverse (cross-beam) resolution As a rule of thumb, the actual –3dB opening angle of a well designed Cassegrain antenna is very close to  -3dB = /D At a distance R, this corresponds to a transverse -3dB resolution of r t = R  For the EISCAT UHFat R = 100 km: r t (100km) = 1.22 km

28. EISCAT Antennas: Aperture Area and Gain A result from antenna theory: G = 4  A/ 2 If the antenna aperture is circular with diameter = D, the maximum gain is G max =  2 D 2 / 2 The diameter of an EISCAT UHF antenna is 32 m. Operating at 930 MHz, the maximum gain becomes G max (UHF) = 97260x or 49.88 dBi So when transmitting through this antenna, the power density in the far field is almost 10 5 times the isotropic power density- but at the same time we illuminate only 10 -5 as many electrons, so the total scattered power doesn’t increase ! But: 1)A large aperture area picks up more scattered signal on receive 2)Higher gain translates into better angular resolution

29. EISCAT Antennas UHF antennas –32-meter fully steerable Cassegrain dishes –Wheel-on-track az drive, rack-and-pinion elevation drive –Max. rotation 1.5 turns in azimuth, 0-90 o in elevation –Max. angular speed 1.2 o /second both axes ESR 32-meter antenna –Fully steerable Cassegrain dish –Rack-and-pinion drive in az and el, angular speed up to 3 o /second –Max. rotation 1.5 turns in azimuth, 0-180 o in elevation ESR 42-meter antenna –Fixed Cassegrain –Pointing along tangent to local field line @ 300 km –Feed adjustable to follow the secular variation in field until > 2007 VHF antenna –40 x 120 m parabolic trough –Can be run as two independent, electrically steerable arrays –Elevation range (15 – 90) o –Computer control presently disabled

30. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution: the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

31. Antenna Feed and Receiver Front End: Polariser, Receiver Protector, Noise Injection and Preamplifier The POLARISER converts the high power TE 10 WG mode into a RHC mode that is radiated by the antenna. On receive, the LHC polarised scatter signal is converted into a TE 10 mode that leaves the polariser through its other port and enters the RECEIVER PROTECTOR (RXP), which is a R/C controlled switch that isolates the receiver from the transmitter by more than - 90 dB when the transmitter pulses. In the receive state it is almost transparent (loss  -0.2 dB). The received signal next passes through a NOISE INJECTION coupler. Noise can be injected on command from the R/C for calibration purposes. Finally, the scatter signal enters the PREAMPLIFIER, a low-noise GaAsFET or HEMT amplifier that raises the signal level by  35 dB before sending it on to the first mixer unit.

32. The ESR Antenna Feed and RX Front End

33. How the receiver processes the signal First mixer Second mixer A/D conversion Conversion to baseband ANALOG DIGITAL Low-pass filtering and decimation

34. VHF ReceiverT-UHF ReceiverESR Receivers

35. EISCAT VHF RX Spectrum Windows 214 224 234 MHz Window I Window II Window III Window IV Klystron passband

36. Remote UHF RX Front End Note: To obtain optimum SNR, the polariser must be set to match the polarisation state of the received wave. As this is a function of the pointing geometry (which changes only when the antenna is moved – timing macroscale), the polariser control is included in the EROS antenna control software package. Polariser phase and amplitude offsets are site-specific and can vary over time – beware !

37. Review of the Radar Hardware A generic radar system Controlling the radar: –timing microscale: the Radar Controller –timing macroscale: EROS RF spectrum issues and problems Generating the RF waveform: the Exciters Raising the power level: the Transmitter Power Amplifiers Radiating the high power signal, collecting the scattered energy and providing transverse resolution: the Antennas Recovering the scattered signal: the Receivers Digitising and processing the signals: the Digital Back-Ends and the Process Computers

39. Block Diagram of Channel Board

40. Tromsø UHF VME crate

41. Process Computers and Data Transfer Channel boardsCPU-50 –viaVME backplane (max. 50 MByte/s total) CPU-50 always controls the reading out of data from the channel boards. Depending on the application, it may also process the data further before sending it on, thus the amount of data transferred out of the CPU-50 may be different from that read off the channel boards: CPU-50SparcServer (t45001 etc.) –viap-t-p Ethernet (max. 10/37 MByte/s) The Ethernet link is the bottleneck in most applications !

42. How Hardware Restricts Experiment Design Radar Controller –program memory size limits the complexity of coded experiments Spectrum –not all sites can receive all UHF frequencies reliably –GSM base carriers above 935 MHz make UPL measurements impossible at all UHF sites –in-band interference at VHF complicates full PL measurements TX –Klystron bandwidth limits altitude resolution –Max. modulator repetition rate limits resolution in P-to-P RX –RXP turnaround time/external clutter govern low altitude limit –limited number of channel boards restricts the number of simultaneous plasma line frequencies –buffer memory size limits the complexity of coded experiments Data transfer –limited transfer speed on dedicated Ethernet can be a bottleneck in raw-data-taking type experiments