1. 17/04/2019
Future VLBI systems
Tasso Tzioumis| Facilities Program Director– Technologies for Radio Astronomy
CSIRO Astronomy and Space Science
IVTW November 2018
ATSC May 2016

2. Jupiter and Io as seen from Earth
Why VLBI?  Resolution
Atmosphere gives 1" limit without corrections which are easiest in radio
Jupiter and Io as seen from Earth
1 arcmin arcsec arcsec arcsec
Simulated with Galileo photo
IVTW November 14, 2018

3. Essential Features of VLBI systems – (issues)
Widespread antennas – Baselines 100s  1000s km
 Precise and stable timing critical.
Synchronisation requires independent very stable reference clocks  H-masers
Raw data samples required for Correlation of antenna signals
Transmit baseband / voltage data  Extremely large data rates
Mitigation: Only need 2-bit samples
Data transport – internet changed issue  e-VLBI – real-time
Mass storage disk systems available cheaply.
Disk transport rare (remote antennas). Mainly on internet. Delayed or real time.
Data Format Compatibility  mitigated with software correlators
Simple to re-format on-the-fly OR ingest multiple formats OR use standard VDIF format
Software correlators freely available and supported (e.g. DiFX)
IVTW November 14, 2018

4. Current VLBI systems ***Are custom VLBI systems always necessary??!!
Custom VLBI systems are implemented (needed?)
Feeds/receivers: mainly standard telescope systems (but see VGOS systems for IVS)
RF/IF chain: Standard till IF (e.g. IF of GHz)
Digitisation: Custom. Special h/w required.
Digital processing: Custom h/w and firmware (DBE; DBBC; KVN/VSOP; LBA)
FPGA based e.g. ROACH
Data format: Custom. Many legacy formats. New standard VDIF.
Mitigated by disks and internet access.
Correlation: Custom. Software correlators.
Mitigated by freely available correlators on HPC clusters (e.g. DiFX, STXC)
 VLBI as special telescope sub-system
Special and time-consuming setups. Difficult and disruptive to schedule.
Only for “vlbi gurus”.
***Are custom VLBI systems always necessary??!!
IVTW November 14, 2018

5. (H-maser; GPS; Freq references; UTC; ….)
RF/IF
System
(LNAs,
Cryo,
RF,
LO/IF)
ADC system
(Sampling,
Digitisation)
Digital system
(FPGAs)
Channels
Packets
Digital switch .
Antenna systems
(Drives,
Feeds)
Compute cluster
(CPUs,
GPUs) RF IF
Data
streams
packets
Timing systems
(H-maser; GPS; Freq references; UTC; ….)
Outputs
Radio telescope signal flow block diagram
Universal backend
COTS systems
RFSOC
IVTW November 14, 2018

6. VLBI with modern telescope Receivers-Backends-GPUs
Problem faced with Parkes UltraWideband (UWB) system
Digitisation at Receiver in Focus cabin.
RF/IF not easily available for VLBI system.
 Need different/new VLBI systems???
BUT
New telescope systems already produce data-packets in GPU cluster!!
Voltage data samples transmitted to GPUs – e.g. Pulsar timing
Samples tagged with precise timing (nsec) for pulsar timing!
H-maser stability already for all systems. Also needed for pulsars.
 Data in GPUs already VLBI-ready!! (Baseband & Timing) !!!
Can use GPUs to channelise to VLBI channels
Reformat to 2-bit and VDIF in software.
Already connected to Disks and Internet.
NO need for Special VLBI h/w or Firmware !!
IVTW November 14, 2018

7. Huge advantages of new Systems
Can access any Receiver or Band the telescope supports
Potential for wide-band VLBI
No need for special setups taking hours….
Just another telescope back-end mode.
Easier to schedule and run.
Cost reductions – no costly special VLBI systems
No special maintenance. No VLBI h/w experts.
Observations as per “normal” telescope operations.
Natural fit to modern telescope subsystems (e.g UWB; PAF)
Can adopt for current telescopes when modern receivers installed
E.g. Parkes adopts UWB back-end for ALL receivers!
IVTW November 14, 2018

8. Characteristics of New systems
Front-end (Feed; RF/IF/LO) – Any of the receivers available on antenna
Can use new UWB systems to cover large band ranges (6:1)
UWL GHz already at Parkes. Tsys ~20K;
Planned: UWH 4-24 GHz
Alternative: UWB-mid 4-16 GHz & UWB-H GHz; share IF conversion.
Use Multibeam or PAF feeds
Universal back-end: ADCs and FPGAs
Already available: ADCs at 4 Gsps and 6 Gsps
Just available: RFSOC from Xilinx (16 ADCs+FPGAs) on 1 chip!!
Commercial of the self (COTS) systems: Switches and GPUs
Massive switches are getting cheap
GPU clusters already available
Astronomy software on GPUs (e.g Pulsars) available free in astro community (e.g Parkes package)
Very large community investment in software
IVTW November 14, 2018

9. Summary and Conclusions
New telescope systems already produce data suitable for VLBI !!
Synergies with other fields (e.g. pulsar observations and timing)
Can leverage Receiver developments (eg UWB) to expand Freq and BW range of VLBI
Rapid developments in ADCs+FPGAs  compact and simpler digital systems
 NO need to always have “special” VLBI h/w and s/w at the telescopes
Simpler maintenance
Simpler setups and schedules.
Part of normal telescope operation – just another “mode”
Opportunity to move VLBI into the mainstream of telescope operations
 More accessibility to VLBI technique
IVTW November 14, 2018

10. CSIRO Astronomy and Space Science Tasso Tzioumis
Facilities Program Director– Technologies for Radio Astronomy
CSIRO Astronomy and Space Science