eleventh synthesis imaging workshop socorro, june 10-17, 2008 the vlba sensitivity upgrade craig...

Eleventh Synthesis Imaging Workshop

Socorro, June 10-17, 2008

THE VLBA SENSITIVITY UPGRADE

Craig Walker, NRAO

2

Eleventh Synthesis Imaging Workshop, June 10-17, 2008

CONTEXT

• The VLBA is based on 20 year old technology• Only limited new capabilities have been added

– Switched from tape to disk• Improved operations, but did not increase capability

– Added 86 GHz: High resolution, but low performance

• Technology advances now enable a significant increase in scientific capabilities at modest cost– Improved signal processing with Field Programmable Gate Arrays

(FPGA)

– Wider bandwidth data recorders

– Computers are now far more capable

– Better low noise amplifiers at high frequencies

• Major funding, like eVLA, is not now available for VLBA

3


UPGRADE PROJECT GOALS

• Bandwidth increase to improve the continuum sensitivity in all bands– Allow observation of weaker sources - hence more sources– Image lower level structures in bright sources– Image polarized structure of weaker features– Allow use of closer calibrators for better phase referencing

• In-beam calibrators more likely• Parallax and proper motions of more objects

– Some of the highest impact science now being done on the VLBA

• Improve the 22 GHz receivers– Increase the sensitivity for both continuum and spectral line projects– Primary goal to help the key project to measure Ho using H2O

megamasers

• These goals are at least partially funded. Costs moderate.

4


TECHNICAL OVERVIEW

• Increase bandwidth to 1024 MHz (4 Gbps with 2 bit samples)– Current bandwidth 32 MHz sustained and 128 MHz peak– Sensitivity increase by factor 5.7 compared to 32 MHz– New digital backend (samplers, signal processing)– New recording system– New software correlator– All systems will be available in 2009. Sooner for the correlator– The major cost is disk supply. Sets time scale for sustained use (2011)

• Upgrade 22 GHz low noise amplifiers– Sensitivity increased by a factor of 1.6 at 22.2 GHz– New standard band at 23.8 GHz for more sensitivity– MPIfR funding - project complete

5


NEW VLBA SIGNAL PROCESSING

Timing module

IF Distributor

VDBE (1)

VDBE (2)

MARK5 C

MARK5 A

IF Distributor

10GE (2) Data

Ethernetswitch

1GE (2) M&C

T-1 Network connection

10GE Data

• Utilize the 500 MHz IF signals that already exist– No new IF/LO electronics required

• Baseband converters, samplers, and formatter to be replaced with VDBE (VLBA Digital BackEnd)

• Recording system upgrade to Mark5C (4Gbps)

6


VDBE

• Samples, filters, and formats data• Two at each VLBA site• Each VDBE samples 2 analog IF signals of 500 MHz bandwidth (8 bit)

– Sample clock is 1024 MHz

• Digital filters form the baseband channels and resample to 1 or 2 bits– Wide range of bandwidths possible

• Data formatted and reordered for one channel per Ethernet frame• Data sent by 10GigE Ethernet to Mark5C or other media• Can send data to the EVLA correlator (Pie Town link) by fiber• The VDBE also obtains calibration data like Tsys and pulse cal • VDBE project is funded and should be complete in early 2009

7


VDBE FILTERS

• VDBE uses CASPER Lab ROACH board– Contains large FPGA

• Polyphase filter option to be developed by Haystack– Restricted frequency settings, but good for continuum

• Digital Down Converter (DDC) option to be developed at NRAO– More flexible frequencies and bandwidths

– Mainly for spectral line observations

• Both will provide very stable and well defined bandpasses• Can switch options easily• Could have other options for FPGA personality

8


VBDE BLOCK DIAGRAM

1.024 GHz Clock 1 PPS32 MHz Clock

CX4/4x

Z-DOK

Z-DOK

V5LX110TFilter Bank

and QDR Memory

4 passbands

EthernetData

IBOB-2 V2

CX4/4x

CX4/4x

CX4/4x

10/100/1000T

PPC 440 EPxComputer

AGC &

AAL

LVDS GPIO + CLK

USB2

PCI

RS232

8 bit Sampler

SMA

SMASMASMA

120 VacPower supplies

NRAO VDBE V 1

Ethernet Control100 Mbit

AGC &

AAL

8 bit Sampler

SMA

TTL GPIO

•

9


ROACH BOARD (Was iBOB2)

• UCB CASPER Lab product– Broadly useful for radio

astronomy signal processing

• Hardware designed by NRAO, KAT, and CASPER

• Small changes needed

10


MARK5C RECORDING SYSTEM

• Disk based, 4 Gbps recording system– Developed by Conduant, Haystack, NRAO– Uses existing Amazon card like Mark5B+– Modules compatible with Mark5A/B – Earlier Mark5’s can be upgraded

• Data input by 10GigE Ethernet– Only cares about packets - not specific format– Data files will appear as standard Linux files on playback

• Mainly meant to be used with software correlators

– Playback through PCI bus– Requires 2 modules for full bandwidth

Mark5A

11


MARK5C STATUS

• NRAO has issued purchase order for 3 prototypes– Price high to include development

• Prototype hardware delivery – Most of the hardware is in house being used for software

correlator testing– The daughter board that receives 10GigE is expected soon

• Software may take until early 2009• Deployment funds not yet identified

12


THE VLBA SOFTWARE CORRELATOR

• DiFX written by Adam Deller at Swinburne

• VLBA integration by Walter Brisken

• Current cluster has 5 units each with 2 motherboards each with dual quad-core cpus (80 cores)

• Benchmark: 420 Mbps for 8 basebands, 2 bit/sample, parallel hands only, 10 stations, 256 spectral channels/baseband

• Will likely replace the hardware correlator during 2008

• To be expanded as needed– Will cost less than disks for any bit rate

• Under test– Phases agree with hardware correlator to

under 1 degree in spacecraft test

13


SOFTWARE CORRELATOR ADVANTAGES

• Fast development• Highly flexible

– No fixed limits on numbers of stations, bandwidth, number of channels etc.

– It just slows down as you ask for more– Relatively easy to add capabilities

• Hardware is commodity servers– Low cost, fast deployment, easily upgraded– Can purchase most after the software is finished

• Works with many types of recording media• The correlator hardware will cost less than the disks for any

sustained bit rate• Community development (Australia, NRAO, USNO, MPIfR …)• BUT: Still too expensive and uses too much power for larger

correlators like EVLA and ALMA

14


EXAMPLE SOFTWARE CORRELATOR RESULT

• Asteroid Radar• Project of Caltech graduate student Michael Busch• Narrow-band VLBI

– Radar signal bandwidth = 40 Hz

– Correlated channels (below) at 122 Hz.

15


22 GHz AMPLIFIER UPGRADE

• Replaced Low Noise Amplifiers (LNA) with modern devices– Amplifiers from the NRAO Central Development Lab

• Other minor improvements to receivers and to subreflector focus and rotation settings

• Project finished in January 2008• Average SEFD results for inter measurements:

– SEFD = System Equivalent Flux Density, a good measure of sensitivity– Before upgrade (22.2 GHz): 815 Jy– After upgrade (22.2 GHz): 502 Jy

• Sensitivity increased by factor of 1.6

– After upgrade (23.8 GHz): 441 Jy• Sensitivity higher than 22 GHz before by factor of 1.8• Away from the center of the atmospheric water line

– Now recommend continuum band is at 23.8 GHz

• Funded by MPIfR

16


Cryo-3LNA(new)

Cold Plate Saddle Wings

Folded Out

To ~ 45°

Cal Coupler

Waveguide-Coax

Adapter(new)

Polarizer Attaches Here

Cryo 3 LNA Installation

Waveguide-Coax

Adapter(new)

17


BEFORE AND AFTER RECEIVER TEMPERATURES

VLBA K-Band LCP Baselines (F109-07 not represented)

0

20

40

60

80

100

120

140

21.421.521.621.721.821.922.022.122.222.322.422.522.622.722.822.923.023.123.223.323.423.523.623.723.823.924.024.124.224.324.4Fsky (GHz)

Temp (K)

L01b L02b L03b L05b L06b L08b L11b R09b L04b

VLBA K-Band LCP Upgraded

0

5

10

15

20

25

30

35

40

45

21.4 21.6 21.8 22.0 22.2 22.4 22.6 22.8 23.0 23.2 23.4 23.6 23.8 24.0 24.2 24.4Fsky (GHz)

Temp (K)

L01u L02u L03u L05u L06u L07u L08u L09u L11u L04u

VLBA K-Band RCP Upgraded

0

5

10

15

20

25

30

35

40

21.421.521.621.721.821.922.022.122.222.322.422.522.622.722.822.923.023.123.223.323.423.523.623.723.823.924.024.124.224.324.4Fsky (GHz)

Temp (K)

R01u R02u R03u R05u R06u R07u R08u R09u R11u R04u

VLBA K-Band RCP Baselines (SN07 not represented)

0

20

40

60

80

100

120

140

21.421.521.621.721.821.922.022.122.222.322.422.522.622.722.822.923.023.123.223.323.423.523.623.723.823.924.024.124.224.324.4Fsky (GHz)

Temp (K)

R01b R02b R03b R05b R06b R08b R09b R11b R04b R09b R04b

140K

40K

40K

21.4 to 24.4 GHz

18


22 GHz UPGRADE ZENITH SEFD

0

200

400

600

800

1000

1200

1400

SC HN NL FD LA PT KP OV BR MK

STATION

SEFD (Jy)

Before 22.2 GHz RCPBefore 22.2 GHz LCPAfter 22.2 GHz RCPAfter 22.2 GHz LCPAfter 23.8 GHz RCPAfter 23.8 GHz LCP

19


LONG TERM

• Prime science goal: <5 microarcsec astrometry (10 as routine)• Bandwidth increase to 4 GHz (16 Gbps)

– Requires new IF/LO system on antenna (expensive)

• Upgrade 43 GHz amplifiers• EVLA 4-8 GHz receiver

– New maser lines and in the sweet spot for phase referencing

• Wide band 22 GHz (18-26 GHz eVLA receiver)• Water vapor radiometers for phase calibration• Improve surfaces for 86 GHz

– Especially fix Hancock surface and add receiver• Data transfer by fiber networks (eVLBI) (required by paying customer?)

• Ka band (26-40 GHz) receiver (required by paying customer?)

• Added collecting area (SKA related?)• All of these require funding

20


END OF PRESENTATION

eleventh synthesis imaging workshop socorro, june 10-17, 2008 the vlba sensitivity upgrade craig...

Documents