Current LBA Developments

Document Sample
Current LBA Developments Powered By Docstoc
					Current LBA Developments

       Chris Phillips

       Disk Based Recording
• Thanks to initial work by Uni of Tasmania
  and Swinburn Uni, the LBA has disk based
  recorders at all Australian stations & NZ
• Based on Metsähovi “VSIB” digital input
  – “PCEVN”
• Have just ordered 10 systems for
  uniformity across LBA
            System Setup
• LBA DAS for sampling and digital filtering
  – Produces up to two 16 MHz bands for 2 IFs
  – 256 Mbps data rate
  – Parkes, Mopra and ATCA have 2 DAS (512 Mbps)
• MRO “VSIC” converts S2 input data into
  VSI-H format
• MRO “VSIB” PCI card used for data input
  into standard PC
• Data recorded onto disk using normal Linux
         Recorder PC setup
• Dual processor (AMD64 )Tyan motherboard
• Dual onboard gigabit Ethernet
• Primary disk recording into Apple Xserve
  – 14  400 GB disks in RAID5 configuration
• 4  400 GB removable IDE (PATA) disks in
  software RAID configuration for
  – Running on Promise Ultra133 TX2
• Also 2  400 GB internal SATA disks
  – 4 onboard SATA-II connectors
            Recorder Costs
• PC with 800 GB disk space
  – €2410 ($A3890)
• VSIB+VSIC input cards
  – €1090
• Apple Xserve RAID with 5.6 TB disk space
  – €9900 ($A1600)
• 4x400 GB IDE PATA disks
  – €1014 ($A1640)
          Why Not Mark5?
• Mark 5a not compatible with LBA DAS
  – MRO systems engineered to work with S2 data
• MRO system better suited to software
  – Data recorded using normal Linux filesystems
• Cost is similar to Mark5
  – Apple Xserve RAID is major component
   Huygens VLBI Experiment
• Intended to be Mark5
  – Not compatible with LBA (Parkes, ATCA,
    Mopra, Ceduna)
• Installed second DAS for experiment
• Produced special cable to record data from
  2 DAS onto one disk recorder
     Real Time Fringe Testing
• Suite of networked Perl scripts allow
  almost realtime interactive fringe testing
• 0.5-1 seconds of data transferred to
  Swinburne to correlate in software
  – Limited by slow networks,
     • 512 kbps for ATCA, Parkes
     • 33kbps modem for Hobart!
• Can get fringes within 30sec
  – Usually a couple of minutes
• Routinely run for all LBA experiments
         Observations to Date
• First full LBA test experiment in April04
    – Including Hart, VLBA, Kashima (Mk5, K5)
•   Also Nov04, May05
•   Tests to NZ
•   Huygens experiment and Huygens tests
•   High spectral resolution of OH masers
    – 4 MHz, 16000 channels
       Disk VLBI Availability
• Disk based recording added as option for
  OCT05 observing session
  – Shared risk basis
• Data recorded at rates up to 512 Mbps
  – Possibly up to 1 Gbps
• Will be correlated using Swinburne
  software correlator
• Suitable for sensitive observations, high
  spectral resolution, pulsar coherence
  dedispersion etc
       Future Developments
• MRO systems are not compatible with
  ATNF S2 correlator
  – Delay tracking performed by S2 playbacks
• All disk & eVLBI correlation using
  Swinburne software correlator
  – May be able to do disk  S2 copying
• On-going project to record 2x64 MHz
  bands from DAS (512 Mbps)
  – Currently problem with clocking
• 2x16 MHz 8bit (or 10bit)
• Recorders equipped with dual onboard
  gigabit Ethernet
• Suitable for 512 Mbps transfer
• Initially record and transmit, then remote
• Will experiment with realtime software
  – Initially single baseline
  – Possibly to Uni of Western Australia
              Long Term Plan
• New correlator being developed for ATCA
  – Based on FPGA digital filterbanks
• Plan to use as VLBI correlator
  – Should handle data rates up to 64 Gbps
     • 2 GHz, 2 Pol, 8 bit

Shared By: