Pulsar 2007 Presentation

Transcript

Decade bandwidth receivers for extremely high time resolution pulsar observations Glenn Jones Hamdi Mani Sander Weinreb Electrical Engineering Department California Institute of Technology Pasadena, California jones_gl AT caltech.edu sweinreb AT caltech.edu Outline Decade-band front end work at Caltech
The GAVRT 34m telescope
Ultra-wideband pulsar processor and spectrometer for GAVRT
Candidate Decade BW Cryogenically Cooled Feeds

Commercial feed designed for antenna ranges Chalmers “Eleven” Feed from Per Simon Kildal Caltech-Developed Cryogenic LNA’s Caltech EE has Delivered 160 LNA’s to Other Research Centers During the Past 4 Years This does not include LNA’s for the 350 element Allen Telescope Array or the 64 element U. of Arizona Submillimeter Camera 4-12GHz LNA #82D at 12K MMIC: WBA13, CIT1 4254-065 , R8C2 Bias: Vd=1.2V, Id=20mA, Vg1=2.33V, Vg2=2.33V 0.5 to 11 GHz, Tn < 5K 18 45 16 40 14 35 Noise Temp(K) 4 Models, @ 12K 50 12 4 to 14 GHz, Tn < 8K 6 to 20 GHz, Tn < 12K 11 to 34 GHz, Tn < 20K 30 Noise Temperature (K) Gain (dB) 10 25 8 20 6 15 4 10 2 5 0 0 0 1 2 3 4 5 6 7 8 9 Frequency (GHz) 10 11 12 13 14 15 Gain, dB 20 GAVRT: Goldstone Apple Valley Radio Telescope Existing subreflector Rotatable tertiary Operated by the Lewis Center for Educational Research Scientists receive large amounts of observing time in exchange for working with educators to develop primary or secondary school lesson plans to involve the students. Cryogenic front-end Receiver box on surface at vertex GAVRT RF Specs




High band cryogenic feed covering 2-14 GHz (~30K noise temp) Scaled low band uncooled feed covering 0.5-4 GHz (hi/low crossover still not pinned down) Polarization switching radiometer. 4 independent IF processors can each downconvert up to a 2GHz subband. LOs can be stepped in 50ms to rapidly cover whole band. First engineering light end of 2007 First science end of 2008 GAVRT General Purpose Digital Backend


Goal: An FPGA based system that can be reconfigured for continuum, spectroscopic, single pulse/transient, and pulsar timing observations. 4 channels x 2 pols x 2 GHz BW  16 GHz to be digitized. Plan to use the Berkeley Wireless Research Center BEE2 and iBOB FPGA processing boards, and software suite. Two ADC’s and IBOB Board Analog data is digitized using an Atmel AT84AD001B dual 8-bit 1Gsample/s ADC. The ADC can be driven with either single-ended or differential inputs, and can digitize either 2 streams at 1Gsample/s or a single stream at 2Gsample/s. The board is designed to mate directly to an IBOB board for highspeed serial data I/O. Bee2 - Five FPGA Processor 5 Virtex II Pro FPGAs 20 GB DDR2 RAM 18 x 10 Gb Ethernet ports Implementation Strategy The GAVRT telescope is a build-to-cost project. As such, it is imperative to meet the minimum design requirements before tackling additional functionality
Fortunately, the minimum requirements are very basic and can be easily met, even with just one iBOB board.
Possible block diagram Computational Challenge of Real-time Coherent dedispersion: The length of the filter The ISM transfer function is : (1/16Hz)*16Gsps = H ( f ) = e jπD / f 1 billion point impulse response! so the phase term is : φ ( f ) = πD / f ⇒ ∆φ = − πD 2 ∆f f To avoid phase ambiguity, we need However, if we process each subband separately, we need only (1/16Hz)*4Gsps = 250 million (still a lot) Can also further break up subbands f2 (2 ⋅109 ) 2 ∆f < ≈ = 16Hz! −16 D (60) /( 2.41⋅10 ) Another trick = Each small filter only needs to be 1/nth as long. Calibration


It will be necessary to know the relative phase between sub-bands to properly dedisperse the signal. Thus we are planning to inject a calibration signal, perhaps from a comb generator. It may be necessary to have a separate transmitting antenna for the calibration to correct for changes in the phase center of the feed versus frequency. Questions for the Pulsar community Is there any desire to continuously record 8 GHz of bandwidth, or just trigger on interesting events?
How to best take advantage of the large bandwidth while still recording at a reasonable data rate for pulsar timing? What sort of preprocessing is desirable for pulsar timing observations?
Acknowledgements
Sander Weinreb & Hamdi Mani (Caltech) 
Tom Kuiper (JPL) 
developed the cryogenic wideband front end responsible for GAVRT signal processing Lewis Center for Educational Research Thank You! Extra slides Data rates and cost of various media Type Data Rate Cost/GB Cost/(1 GB/s transfer) RAM >4GB/s <$100 <$100 ~$0.30 ~$840 Hard Disk ~60MB/s (16.7 drives * $50 each) Flash ~10MB/s $5? ~$500 Large Overhead Motivation: Why so much bandwidth?
Bottom Line: Allows studying singlepulse phenomena (giant pulses, nanostructure) with the highest time resolution. Noise Temperatures of LIndgren Feed and Prototype of Chalmers Feed Lingren quadridge/Vivalid feed measured with ambient/sky Y factor at Goldstone Chalmers feed with cooled Krytar balun measured with ambient/LN2 Y factor at Caltech Dec 17, 2006 200 180 160 140 120 100 80 60 40 20 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Radiometer Block Diagram TEMPERATURE CONTROLLED AT 40C CRYOGENICS DEWAR, 4 TO 14 GHZ POLARIZATION SELECTOR +20dB 30 dB CPLR 2-14 LNA -5 dB BZP520GB1 DUAL LINEAR WIDEBAND FEED TRANSER SWITCH PIN DIODE 30 dB CPLR 2-14 LNA 4-WAY PWR DVIDR POST AMP SWITCHED CIRCULAR POLARIZER HYBRID TO OTHER IF CONVERTERS DUAL-CHANNEL IF CONVERTER DUAL-CHANNEL IF CONVERTER DUAL-CHANNEL IF CONVERTER DUAL-CHANNEL IF CONVERTER -8 d/b SIGNAL DISTRIBUTOR 4-WAY PWR DVIDR POST AMP MATRIX SWITCH, 16:8 CONNECTS 1 OF 16 INPUTS TO 1 OF 8 OUTPUTS. (1 OF 4) INPUT: 0.5 TO 18 GHZ, DUAL-LINEAR, FROM EITHER FROM EITHER FEED/LNA OUTPUT THROUGH 4 FIBERS CARRYING TWO POLARIZATIONS, I AND Q PHASE, OR UPPER AND LOWER SIDEBAND 0 TO 2GHz OVER FIBER EACH IF CONVERTER CAN SELECT ANY CENTER FREQUENCY AND BANDWIDTHS RANGING FROM 2 GHZ TO 100 MHZ. SEE DETAILED BLOCK DIAGRAM 100 MHZ +7dB CRYOGENICS CONTROL TEMP SENSORS VAC SENSORS LNB REFRIG CONTROL HIGH BAND NOISE DIODE LNA BIAS 8 DRAIN AND 8 GATE SUP -5 dB POST AMP DUAL LINEAR WIDEBAND FEED TRANSER SWITCH PIN DIODE 30 dB CPLR 2-14 LNA MONITOR, CONTROL AND POWER DISTRIBUTION CONTROL DRIVERS MICROPROCESSOR FIBER OPTIC TRANSCEIVER MUX A/D 16- BIT DC POWER BUSS +/-15V, +5v POLARIZATION SELECTOR 1.2 to 4GHz Feed & LNA 2-14 LNA MOTHERBOARD NOISE DIODE DRIVER LOW BAND NOISE DIODE PUMP, VALVE, HEATER CONTROL 30 dB CPLR NCAL POST AMP 4-WAY PWR DVIDR SWITCHED CIRCULAR POLARIZER HYBRID M/C TEMPERATURE CONTROL AND DC POWER 117 VAC 4-WAY PWR DVIDR THERMOELECTRIC CONTROL HEAT/COOL +24V POWER TEMP CONTROL +15V, -15V +5V POWER MOTHEBOARD 9 AMPS OUTPUTS DRIVE PHOTONICS TRANSMITTERS TO 8 FIBERS IF Converter and Baseband Processor DUAL-CHANNEL IF CONVERTER (1 OF 4) 0.5 TO 18 GHZ UPCONVERT MIXER PRE AMP ACTIVE X2 +12 dBm -6dB BANDPASS FILTER 22/2.GHZ +27dB +15dB ZX60-6013E +9dB FILTER 21.3/0.4GHZ 10 - 1000 120 - 520 100 - 1000 QUAD HYBRID LNA + I/Q MIXER 270 - 370 10 - 1000 500 - 1000 HMC571LC5 120 - 520 ADA-1020 -2 TO -33.5dB 500 - 1000 BANDPASS +13dBm BZP518GB1 Mech Latch Switch POWER METER 21 TO 23 GHZ NF<6db MMC560LM3 or M9-0942LN P1DB.> 0dBm -1dB +10dB -13dB INPUTS FROM LNA'S BASEBAND ANALOG PROCESSOR (1 OF 4) +4dBm +27dB 6-BIT STEP ATTEN IF MON IF MON 6-BIT STEP ATTEN HMC542 270 - 370 POWER METER +7dBm 19 GHz LOW PASS FILTER ACTIVE X2 ADA-0512 INPUTS FROM LNA'S PRE AMP +8dBm POWER DIVIDER 11 - 19 GHz POWER DIVIDER ACTIVE X2 +27dB UPCONVERT MIXER POWER METER -2 TO -33.5dB 500 - 1000 BANDPASS FILTER 22/2.GHZ 10 - 1000 120 - 520 100 - 1000 QUAD HYBRID LNA + I/Q MIXER 270 - 370 BANDPASS 10 - 1000 FILTER 21.3/0.4GHZ 500 - 1000 120 - 520 SYNTHESIZER 5.5 TO 9.5 GHz 270 - 370 100 MHz REF REF LO DISTRIBUTOR 11.0 GHz PLO 100 MHz 8-WAY PD 100 MHz CLEAN-UP PLO FIBER OPTIC RECEIVER 100 MHz REF +27dB 6-BIT STEP ATTEN IF MON IF MON 6-BIT STEP ATTEN HMC542 POWER METER