Day1_Session1_Notes - david-macmahon/wiki_convert_test GitHub Wiki

Who wants to do what:
=====================

Don Backer: ATA 100MHz complex signal dual-pol into pulsar machine
            Increase bandwidth capability on ASP/GASP.
Laura Spitler: Wants to know who is doing what
Peter McMahon: Here for a year at full-time
John Ford: NRAO pulsar machine; spigot replacement first; coherent dedisp later
           Trying to understand CoDeDi, and interface between FPGA and computer cluster
Jason Ray: over the next year ramp up involvement
Tim Ikin: good pulsar back end for Lovell; have a Ser5 board + engineer and hope to put
          that together soon. Effelsberg needs back end for 100Mhz 7-beam receiver;
          Parkes want a digital filterbank for the 300MHz 13-beam multibeam to replace analog;
Paul Prozesky: iBOB-based back end for XDM=1-dish KAT demonstator; interested in seeing what's happening
Randy McCullough: as John; how to implement folding? Have a list of questions for today
                Goal to reestablish as fore-front technologists; foster collaborations with Universities
                Use boards and toolset to farthest extent. Are now committed to CASPER approach.
Glenn Jones: General purpose machine, sprectroscopy should be straight forward; exploring timing
             and data issues; dedispersed trigger;
Joeri van Leeuwen: cf. Don for Berkeley goals; for CASPER, hopes to identify overlapping stages and coordinate who can do what.



Open discussion:
================

John:
  Spigot replacement (as on NRAO Wiki):
    data rate could be 1-3 Gbps; write to 8-node back end;

Don:
  The pocket spectrometer, put two on an iBOB, cross correlate have a Stokes detector
  Where to put the IF?

John: Plan was to use 800-1600.
        PFB FIR on two iBOBs; PFB FFT on BEE2

Don: GASP Midend, 256MHz over 4MHz chunks, 10->1nodes, feed ARTS software
     is close to the VLBI system
     diagram of how ADC+S5+dataservers what will be replace by single iBOB
     ATA: 100MHz band width?
     And of course, absolute time: how to get the 1PPS through your system
     GASP uses a 'trigger' and 'arm' (cf. picture 2)

-->  [audio starts here]

Glenn: Comb filter, because they have 4 bands? Any one else?

Don: A good little project could be to make sure we can read the 1Gbps at full speed, just PC to PC?

Don:  Glen, what's the plan for the GAVRT trigger?
Glenn: that 2iBOB+1BEE on the left is what he was thinking of. Not yet sure on how much BW to do
       the trigger. Maybe 100MHz or so. Possible incoherently dedispersed on the BEE.
       NOt sure if there's need to incodedi? Can you get better time resolution that your sampling time?
Don:  sure, you can subdivide bins. Make sure to have more than 1k bins per pulsar period.
      And at high frequencies you might not have to do codedi anyway.
      If within you incodedi bin one part if brighter than another (cf fig 3) your arrival time will
      vary with scintillation.
Randy: Different projects then want different bit rates, 4,8,16 bits
Don:   And want to mention RFI. Some sensor for samples of > 5 sigma and inject noise/zeros would be new
       and very useful. Possibly before and after the PFB, maybe just blank top digits.