EMBRACE@Nançaysatorchi.net/skads/FranceChina_EMBRACE_20111103.pdf · 15 France-China SKA Workshop,...

1France-China SKA Workshop, Meudon 2-3 November 2011

Preparing for SKAEMBRACE@Nançay

Steve Torchinsky

EMBRACE@NançaySteve Torchinsky

Station de radioastronomie de NançayObservatoire de Paris



Steve Torchinsky

EMBRACE@Nançay



Steve Torchinsky

Electronic MultBeam Radio Astronomy ConcEpt

(EMBRACE) EMBRACE is an AAmid Pathfinder for SKA

Largely funded within EC FP6 Project SKADS (2005-09)

For EMBRACE:

ASTRON: Project Leader, overall architecture, antennas,

industrialization,...

Nançay: Beamformer Chip, Monitoring and Control Software

MPI Bonn and INAF Medicina: design of multiplexing circuits for

RF transport, down conversion, and command/control

Two demonstrators built. One at Westerbork (144 tiles) and one at

Nançay (64 tiles currently)



Steve Torchinsky

EMBRACE@Nançay 4608 Vivaldi antenna elements Single polarization (second polarization antennas are

there for a total of 9216 elements, but only one polarization has a complete signal chain)

4 level hierarchical analog beamforming/signal summing Beamformer chip:

− 4 inputs, 2 outputs (2 independent beams)− 45º phase steps

Analog summing output from 3 beamformer chips Analog summing of 6 inputs = 1 tile (72 elements) 15m cable → Analog summing of 4 inputs = 1 tileset Down conversion → 16 inputs to LOFAR backend



Steve Torchinsky

EMBRACE@Nançay 500 – 1500 MHz

– But high pass filter at 900 MHz to avoid digital television 70 m2 (10.5m X 10.5m) Instantaneous RF band: 100 MHz Maximum instantaneous beam formed:

– 36 MHz x 2 directions (single polarization)

– 186 “beamlets” each of 195.3 kHz bandwidth

– ie. 3 “lanes” for high speed data from RSP

– Can trade off beam width vs. number of beams



Steve Torchinsky

EMBRACE@Nançay



Steve Torchinsky

Hierarchical beamforming

4 elements phasedTogether in theBeamformermicrochip



Steve Torchinsky

Nançay Beamformer Chip



Steve Torchinsky


4 elements phasedTogether in theBeamformermicrochip



Steve Torchinsky


3 beamchips on aHexboard(12 elements)



Steve Torchinsky


6 hexboardsin a tile(72 elements)



Steve Torchinsky


4 tilesin a tileset(288 elements)



Steve Torchinsky

4th stage analog beam forming



Steve Torchinsky

Control and Down Conversion● Last stage analog summing of four

tiles● 2-stage mixing to convert RF down to

150 MHz +/- 50MHz● Ethernet protocol for beamformer

chip parameters and housekeeping● 48V DC● RF + Digital Commands + Power

ALL ON THE SAME COAX!

CDC cards designed by INAF-IRA/MPIfR/ASTRON



Steve Torchinsky

Control & Down Conversion

● CDC cards group 4 tiles together (tileset)● RF and control/command multiplexed



Steve Torchinsky

LOFAR Backend

● Output from CDC goes to LOFAR Receiver Unit (RCU) boards for digitization

● And then to LOFAR Remote Station Processing (RSP) boards for digital beamforming



Steve Torchinsky

System Control and Data

• Enormous flexibility with the dense array– Multi-beam– Instantaneous reconfiguration– Real time calibration– Multiple observing mode possibilities with tradeoff

between bandwidth, number of beams, field of view



Steve Torchinsky

System Control and Data

• Extension of LOFAR Local Control Unit Software– More complex task with EMBRACE especially

regarding beamformer chip control

• Station Control Unit software– Python modules

– User interface simple, but highly adaptive using python scripts

– Statistics saved in FITS files



Steve Torchinsky

Antenna dataFrom RCU

2N x 200 Ms/s12 b.

SeparateSubbands

SelectSubbands

FormBeams

OutputBeams

ComputeSubbandsStatistics

ComputeCross

Correlations

ComputeBeams

Statistics

Correct forCalibration

Nulling ofInterferers

CalculateInitial vector forBeam forming

CalculateCalibration

DetectInterferer

CalculateProjection

Matrices for nulling

Subbandfrequency

Array geometry

Subbandfrequency

Array geometry

Source coordinates Interferers coordinates

Station Control Unit

SubbandTo be processed

Output modeTime stamp

LO1 beam A

LO1 beam B

LO2

Data recordingPost processing

LocalControl

Unit

N cells of combined tiles

2 RF beams

Store Store Store

Tile arraysettings

CalculateTile arraysettings

Source coordinates

ExternalCorrelatorInterface

Mark V recording

EMBRACE Monitoring & Control



Steve Torchinsky

Phase Calibration using GPS

● GPS BIIA-nn: ● Strong carrier at 1227.6MHz● Narrow band

● GPS BIIF-1:● Strong carrier at 1175.6 MHz● Wider band (~7 MHz)



Steve Torchinsky

Without Phase Calibration

● Subband statistics

● 100 MHz band per tileset

● Each tileset pointing differently (or not properly)



Steve Torchinsky

With Phase Calibration

● Tilesets are together● Next step: level equalization



Steve Torchinsky

Drift Scan of GPS BIIF-1

● 2 beams formed● Second beam is 4 time minutes ahead in the path of the satellite



Steve Torchinsky

Drift Scan of Cas-A

● Gaussian main lobe● FWHM 1.476º

● 1.2λ/D = 1.486º



Steve Torchinsky

Drift Scan of Cyg-A



Steve Torchinsky

Pulsar B0329+54

Pulsar B0329+54

EMBRACE@Nançay connected to ARTEMIS backend (courtesy U. Oxford)



Steve Torchinsky

Continuing AAmid Microelectronic Developments

Next generation beamformer chip: 22.5º phase steps Chips delivered and undergoing test



Steve Torchinsky

Aperture Array Integrated Receiver Proposal to French National Funding Agency Successful! 816k€ over 4 years Partners: Nançay (coordinator),Bordeaux, NXP, ASTRON Beamformer chip with time delay instead of phase shifting for

wide bandwidth System-in-Package:

ADC with Serializer LNA with filter

Ultimate goal (timescale of the AIP): AAmid tile with digital output Reduce power consumption, and data transport/command complexity



Steve Torchinsky

Objectifs: Montrer la faisabilité d’une électronique intégrée destinée aux applications de réseaux phasés.Validation du concept par la réalisation d’un mini réseau de démonstration utilisant 12 à 16 antennes élémentaires.

Tuile RF 1m² Amplification, filtrage, beamforming Interface de commande

Micromodule(s) ~qq mm²

Aperture Array Integrated Receiver



Steve Torchinsky

Aperture Array Integrated Receiver



Steve Torchinsky

Science Data Model v2

• Collaboration with Obs Paris - LERMA (F. Viallefond)– Evolution of the ALMA SDM– SDMv1 currently used on ALMA and EVLA– Mathematics of “categories”– Naturally object oriented– XML and Schema– See CALIM2011 for more details:

http://www2.skatelescope.org/indico/conferenceOtherViews.py?view=standard&confId=171



Steve Torchinsky

Science Data Model v2 - Goals

Generalised model for any radio telescope Single dish Aperture synthesis

− (eg. ALMA, EVLA, PdB, ...) VLBI Aperture plane phased arrays (eg. EMBRACE) Focal Plane Arrays

− Feed clusters (eg. Parkes, Arecibo, Effelsberg, ...)− Phased array feeds (eg. ASKAP, APERTIF)



Steve Torchinsky

SDMv2 - Overview Enum

Categorized enumerators PhysQuan (Physical Quantities)

Collections of physical quantities (eg. Temperatures) and measures (eg. QDirections)

Virtual Quantities (eg. Phase_Dir, Staircase for OTF) Values with uncertainties (eg. Position, Length)

SDM Ensemble of Tables

− Relations within tables, and between tables Normalization

− Allows queries in a database Set/subset based



Steve Torchinsky

Science Data Model v2 - components Meta data

eg. definition of a configuration Auxilliary data

eg. instrument diagnostics (temperatures, encoder values, etc)

Data eg. visibilities, autocorrelations

Objects and Types persist and remain accessible



Steve Torchinsky

EMBRACE@Nançay

4608 antenna elements Hierarchical beam forming Organised into tilesets

16 tilesets Each tileset is 4 tiles of 72 elements each “statistics” include

− Cross correlations between tilesets (imaging)

Lots of flexibility regarding bandpass, pointing, number of pointings



Steve Torchinsky

SDMv2 Main Features

Can be easily pruned for a specific instrument

Can be applied to any instrument of any complexity or simplicityeg. future radio telescopes: ASKAP, APERTIF, EMBRACE,

ALMA, EVLA, VLBI, etc

eg. for SKA: It will be composed of multiple technologies, but all can produce data in SDMv2



Steve Torchinsky

SDMv2 - Implementation 3 levels

Meta model: generic parameters (XML Schema) Model: instance of the meta model (XML) Data set: instance of the model (XML + binaries)

Code generation from the model (c++ template) c++ classes XML Schema (Type definitions)

− eg. Dataset<Sdm>, Dataset<Sdm,EMBRACE> Python interface



Steve Torchinsky

EMBRACE@Nançay



Steve Torchinsky

EMBRACE@NançaySteve Torchinsky

Station de radioastronomie de NançayObservatoire de Paris

Date post:	03-Jul-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

EMBRACE@Nançaysatorchi.net/skads/FranceChina_EMBRACE_20111103.pdf · 15 France-China SKA Workshop,...

Documents