CICADA - Green Bank Spectrometer Project

Design Phase Project Kickoff Meeting, September 2, 2010

Attendees

• Randy McCullough
• Jason Ray
• Roger Norrod
• David Frayer
• John Ford
• Anish Roshi
• Amy Shelton
• Patrick Brandt
• Marty Bloss

Agenda

1. Project Organization
   • "Gbsapp" mail exploder established and will be be primary email address for all GB Spectrometer correspondence. Sign up at: http://listmgr.cv.nrao.edu/mailman/listinfo/gbsapp
   • Overall Project Schedule
2. Identify a set date and time for a weekly meeting between NRAO people and the Berkeley group.
3. Review the action items below, add any missing items; identify people to take up the action items and establish time frame for completing the action items.
   • Action items for overall design:
     1. Second (Final?) version of the system specification
     2. Estimation of the FLOPS for the computation to be implemented in the HPC for different spectrometer modes and possible configuration (i.e. if integration for different states are done in the HPC rather than in the hardware, etc).
     3. Complete system design study (hardware and software).
   • Action items for identifying the interface with the spectrometer:
     1. Freq ref signal: freq, sine/square, voltage level, connector
     2. External switching signal: voltage standard, connector
        • BNC connectors
        • 5 volt TTL compatible
        • Switching signals to the spectrometer driven by 74F244
        • Switching signals from the spectrometer drive TTL loads
        • See GBT Spectrometer Drawing 35208L12 and Switching Signal Selector 35201L004 for reference designs
     3. 1 PPS signal : voltage standard, connector
        • BNC connector
        • 50 ohm driver, 5 volt signal.
        • Pulse width unspecified, but a few microseconds wide
     4. Identify a synthesizer, voltage level for ADC, how much amplification needed, power splitter, connector.
     5. Analog part of the spectrometer: design, identify component, connectors. Identify someone in microwave group to lead this work.
     6. Identify the network interface to connect the spectrometer to the GBT computing system; connector
        • Assumed to use existing 1000 base-T for the gigabit netowrks, and either CX-4 or optical SFP+ for the 10 gigabit data links
   • Action items for software
     1. Identify relevant software components available (mainly from GUPPI) and a plan for staff to work on this. (Patrick B. plus Ron DuPlain via UC-Berkeley?)
   • Action items for identifying the requirement for maintenance of the spectrometer:
     1. Signals that should be accessible for monitoring; indication of the presence of signals through LEDs;
     2. Software monitoring: list of items to be available for monitoring
4. Review Risks/Development work and possible action items for the Berkeley group
   1. ROACH1 vs ROACH2 -- Helpful to make decision if we can identify the largest PFB that can be implemented in ROACH 1 for processing the full bandwidth of 1.5 GHz from two polarizations.
   2. Design and implementation of DDC
   3. Overlapping PFB.
5. Identify Project Team
6. Review Project Milestone draft
7. Project Roles (WIP)
   • CICADA Program Manager & Sr. Architect: JohnFord
   • Green Bank Spectrometer Project Scientist: AnishRoshi
   • Green Bank Spectrometer System Architect: AnishRoshi
   • Green Bank Spectrometer Project Manager: MartyBloss
8. Other
   • Project book (example)
   • Need a name for the project. Suggestions so far:
     • Spectralprocessor II
     • GBT spectral and pulsar processor (GBSAPP)
     • GBT Universal Spectral Processing Instrument (GUSPI)

Action Items:

• Spectrometer overall design
  • Anish will lead effort to compile the second iteration of the specification and circulate the draft to the Green Bank scientific staff. Once the scientist's comments have been incorporated, the draft will be circulated to the collaborators for comment. The target date for release of the draft to the collaborators is September 10.
  • A meeting of the software and data-reduction pipeline team members will be held to scope out the magnitude of the computational power that will be required for the different modes
  • Patrick will lead effort to identify which aspects of the GUPPI control architecture should be brought forward into the spectrometer design
    • I'd like to tie down some of the controller features which are on the list for closing out GUPPI (listed at GuppiCloseOutItems), which would make the controller more portable and accessible to casper-based instruments that aren't GUPPI. I intend to find time this month (Sep 2010) to do this. Please touch base with me during this investigation. -- RonDuPlain - 2010-09-03
  • Anish will lead analysis of the maintenance and monitoring requirements
  • Once the above activities are complete, Anish will lead an analysis of the hardware and software trade-offs by the collaboration team and finalize overall design.
• Design and interface details:
  • Baseline definitions for interfaces to the master clock, frequency reference, and switching signal interface agreed upon. (Details included in section 3 above)
• Marty will work with the head of the microwave group to identify an analog engineer to add to the project team
• The Berkeley team will lead the work on matching the technical designs to the specifications and requirements, including identifying and managing the technical risks
• John, Anish, and Marty will construct a development flow chart as a tool to better identify the major project milestones for the project plan
• The project will create a web-based project book along the lines of the 4mm Receiver example. Each chapter will be assigned to an individual who will solicit content and manage the editorial and periodic updates of their chapter.
• The project still lacks a name, the suggested names so far haven't resonated with the team. Post any suggestions to Gbsapp mail exploder
• Marty will post a poll for a periodic overall team meeting