An Integrated Program for GBT Focal Plane Array Development

1
An Integrated Program for GBT Focal Plane Array
Development

• Steven White
• Based on discussions with
• Rick Fisher, John Ford, Matt Morgan, Roger
  Norrod, Kamaljeet Saini, Amy Shelton, Richard
  Prestage, John Webber, and others

2
Summary

• A pathfinder focal plane array (FPA) frontend is
  the necessary first step in an integrated
  development program.
• K-band is the appropriate frequency, for reasons
  which will be outlined.
• We have the necessary resources and expertise (,
  more importantly appropriate staff effort) ready
  and eager to go.
• We need to start now to demonstrate results
  within 2 years (production use in 3 years).

3
Integrated FPA Development Program

• The Science Case is the primary driver for
  frequency selection and detailed specifications.
• Development Program Components
• Frontend (EM Components, Cryogenics, HEMT Amp,
  Calibration, Downconversion)
• Digitization and IF Transmission
• Spectrometer
• Software
• Scientific desires must be tempered by technical
  realities and costs of each component .
• True measure of a successful program for the GBT
• Competitive instrumentation installed on short
  time scales and in production use by external
  observers.

4
Instrument Development Realities

• GBT Spectrometer
• 5 year development time with an additional 5
  year integration time.
• Finally debugged August 2007
• Ka-band Receiver
• Extensive refurbishments required for each of
  three successive observing seasons
• HARP/ACSIS FPA program on JCMT
• HARP 16 pixel 345GHz frontend
• ACSIS 16 x 2 x 1GHz spectrometer backend
• Similar specifications and logistical challenges
  to our proposed system
• 10 year development time
• IF distribution became critical path item
• Commissioning FE/IF/BE simultaneously is
  enormously complicated!
• Solution
• Select challenging but realistic component
  deliverables.
• Develop on 1-2 year timescales.

5
Development Path 1st Step

• Frequency Selection
• ? lt K-band physically too large for Gregorian
  focus (needs beam-forming array at prime focus).
• ? Q, W-band existing IF capacity inadequate
  upgrade path unknown.
• ? Q, W-band telescope performance still under
  development (should be delivered on 2-3 year
  timescales, but not a given).
• ? W-band significant detector and other RD
  development required.
• Where is GBT Unique?

6
K-Band Advantages

• A scientifically exciting but technically
  realistic focal plane array.
• Can do production science with existing IF system
  and spectrometer.
• Parallel and collaborative spectrometer
  developments possible but not critical.
• Realize cost savings and enhanced performance for
  future expansion of instrument.

7
K-Band Advantages

• Develop on 1-2 year timescales
• Integrate each new component into the production
  system to allow immediate, productive astronomy.
• Technical staff available for two years and ready
  to begin.
• Frontend minimal RD effort required.
• Current capability to demonstrate a prototype
  within one year.
• Concurrent software development with a usable
  instrument within 2 years.

8
K-band FPA Deliverables

• Frontend
• Cryogenic Package
• Modular Downconverter
• Modular Noise Calibration
• Mechanical Packaging
• Software
• Package for Engineering and MC
• Package for data analysis

9
Future FPA Developments

• Significantly enhanced spectrometer with this new
  array and existing IF system.
• New digital I.F. distribution system.
• Expanded focal plane array (perhaps at a
  different frequency)
• Phased, expandable software development to match
  hardware capabilities.

10
Spectrometer

• We have a new approach CICADA program
• Configurable Instrument Collaboration for Agile
  Data Acquistion
• Collaboration between GB, CDL, UC Berkeley CASPER
  group, WVU, University of Cincinnati, others
• Initial production instrument pulsar backend
  (Scotts Dream Machine")
• Design of a 2GHz bandwidth spectrometer already
  identified as a deliverable for FY2008
• Construction could commence in FY2009 given
  resources
• Excellent candidate for external funding
• Correct technology area (innovative hardware /
  networking / computing)
• University collaborations
• gt prototype backends are under development
  new Spectrometer development should start in
  Fy2009

11
IF System

• Current analog IF system limits expansion.
• Analog modulators costly with increased
  complexity due to stability requirements.
• Downconversion scheme limits bandwidth
• Digitization at antenna with filtering/compression
  schemes probable solution for IF transmission.
• Commerical technology advancements in digital
  transmission reduces cost and increases
  performance.
• Development in this area is extremely active,
  driven by SKA and SKA-pathfinder initiatives

12
Software

• Similar systems do exist at other telescopes.
• Challenge in software is to transfer approach or
  actual implementation to your specific telescope.
• We will certainly leverage work done elsewhere,
  and collaborate with other groups.
• Using existing GBT Spectrometer makes problem
  much more tractable (many GBT-specifics already
  solved).
• gt Software development is an integral part of
  FPA program addressed early in the project.

13
Development Path Summary

• Processing software issues well understood
• Backend current system can accommodate a modest
  array (at K-band), issues and upgrade path well
  understood.
• I.F. system current system can accommodate
  modest array, upgrades premature at this time
• Frontend Proposed K-band.
• A pathfinder focal plane array (FPA) frontend is
  the necessary first step in an integrated
  development program.

14
K-band FPA Summary

• K-band is the appropriate frequency.
• We have the necessary resources and expertise (,
  more importantly appropriate staff effort) ready
  and eager to go.
• We need to start now to demonstrate results
  within 2 years (production use in 3 years)

15
Backup Slides
Backend Data Transmission System Telescope
Performance Weather
16
Backend Strawman Design

• Strawman Specifications
• 3 GHz analog bandwidth
• 6 GS/s
• 8 bit ADC, 6 bit ENOB
• 183 KHz resolution_at_ 3 GHz bandwidth (16K
  channels)
• 80 millisecond minimum integration time
• 200 MB/s aggregate data rate to disk for 122 IF
  channels

17
Backend Strawman Design

• Technology Approach
• Off the shelf ADC chips
• FPGA DSP hardware
• Off the shelf Data Collection Computer Systems
• Optional locations
• Receiver room
• Equipment room

18
Backend Strawman Design

• Cost
• 20K/pixel sampler and DSP costs(2 channels)
• 2.44M for 61 pixels, scales linearly with
  pixels
• 30K Data collection and storage computers
• 70K data transmission to disks (all DSP at front
  end)
• 6 FTE-years effort (3 people, 2 years)
• Total 3.3 million.

19
Data Transmission Systems

• EVLA and ALMA Data Transmission Systems are not
  appropriate
• GBT needs 4 bit samplers
• EVLA 3 bits
• ALMA 2 bits
• ALMA and EVLA multiplexing too complex and
  expensive for GBT
• gt Advantages in allowing technology to develop.

20
Telescope Performance
21
Azimuth Track Replacement Project

• Field work completed on Monday 3rd September to
  specification, on schedule and within budget
• Now in the process of developing new pointing
  model initial results extremely promising.

Local tilt (pitch) of the antenna as measured by
inclinometers mounted on the elevation axle. Both
large and small scale track effects are
significantly reduced.
Very first observation after outage complete.
Measured pointing offsets (3, 1.5) in (az,el).
22
14GHz half-power track
23
Surface Performance

• OOF technique can easily measure large-scale
  wavefront errors with accuracy 100µm
• Large scale gravitational errors corrected via
  OOF look-up table
• Benign night-time rms
• 350µm
• Efficiencies
• 43 GHz ?S 0.67 ?A 0.47
• 90 GHz ?S 0.2 ?A 0.15
• Now dominated by panel-panel errors (night-time),
  thermal gradients (day-time)

24
Summary Current Performance
25
Weather
26
(excellent)
Atmospheric Absorption and Emission
(mediocre)

• lowers the SNR ? exp(-?) / Tsys

Slide Courtesy Jim Condon
27
Observing Limits (accept/reject)

• Atmospheric efficiency
• Atmospheric stability
• Absolute hour angle
• Zenith angle lt 85 deg
• Tracking flux error lt 10

Atmospheric efficiency
Slide Courtesy Jim Condon
28
Stringency ? 1 / (fraction of time OK)

• Stringency measures the difficulty of
  scheduling an observation on the GBT. It depends
  strongly on the observing frequency ? and also on
  the source elevation at transit.

Slide Courtesy Jim Condon