Transcript SKADS Costing

SKADS Benchmark Scenario
Andrew Faulkner
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Design and Costing Process
Submitted to March ISSC
Memo on SKADS website:
www.skads-eu.org/PDF/SKADS_Benchmark_Scenario
SKADS Benchmark Scenario
Design and Costing
Paul Alexander1, Rosie Bolton1, Andrew Faulkner2, Steve Torchinsky3,
Arnold van Ardenne4, Peter Wilkinson2, Marco de Vos4, Laurens Bakker4,
Simon Garrington2, Georgina Harris2, Tim Ikin2, Mike Jones5, Dion Kant4,
Roshene McCool2 and Parbhu Patel4. On behalf of the SKADS team.
_D+C_v1.0.pdf
Published as SKA Memo 93:
www.skatelescope.org/PDF/memos/Memo_93.pdf
1
University of Cambridge, 2University of Manchester, 3Obs. de Paris, 4ASTRON,
University of Oxford
5
Abstract
The SKADS Benchmark Scenario is an overall SKA concept which aims to
meet as many as possible of the SKA requirements presented in the SKA
reference design. The key element of the system design is the use of aperture
array technology on all baselines below a frequency of 1 GHz which gives a
field of view of 250 square degrees in the key mid-frequency band. At higher
frequencies comparatively low-cost, small (6.1 m) antennas are proposed each
equipped with a single wide-band feed. The detailed design of high frequency
dishes and wide-band, single pixel feeds is likely under a highly
complementary US TDP programme and elsewhere. This Scenario is
presented in detail, concentrating in particular on the design of the key midfrequency aperture array and drawing on the work of other projects for
elements of the system outside of the expertise and scope of the SKADS
project. Detailed costing of the design suggests strongly that the Benchmark
Scenario is a practical and achievable implementation which delivers the
scientific performance for the SKA. The “worst case” cost is estimated to be
€1.91 Billion with an uncertainty of 9% costed for 2011. Further development of
the Benchmark Scenario will include detailed scientific and astronomical
evaluation and simulation together with cost optimisation and cost/performance
tradeoffs.
‘Complete SKA concept’
Strong basis for moving
forward
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Top Frequency....
The cost vs frequency of a close packed aperture array varies as:
Cost ≈ freq 2+ε + infrastructure
cost for ftop of 1.0 GHz < 50% cost for 1.4 GHz
Science expt. requirements
determines ftop as ~1GHz
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
An overall SKA Scenario
AA only
realistic
collector for Scenario:
The
SKADS
Benchmark
• Small dishes lose efficiency at low freq.
<300MHz
• AA gives v. fast survey speeds and can
look for transients
100 - 300MHz
Dipole aperture array
<300 - 1000MHz Close packed aperture array
0.7 - 20+GHz
6-8m, dishes with single wideband feed
• Low cost m-2 and they exist!
• Large natural FoV
• Good at high frequencies
• Capable of mass production
• WBFs could be >20:1 freq range
• Single feeds can be cooled
economically
• Not suitable for arrays
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Outline mid-freq AA spec
Frequency range
300MHz – 1000MHz
Organisation
Close-packed (for now)
Array size
~60m diameter (~300 tiles)
Tsys
<50K
Bandwidth
700MHz
Scan angle
±45° (target ±60°)
Signal digitisation
4-bit
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Filled FoV specification
Low-AA: 200 deg2
1000
Reference Design
SKADS Benchmark
100
Mid AA: 250 deg2
10
FOV (deg 2 )
An AA can tailor
this profile e.g.
1
flat,
λ2 etc.
6m Dish + WBF
3*(1.4/f)2 deg2
0.1
0.01
0.001
0.1
At 1.4GHz ...
Ref:
5 deg2
B’mark: 3 deg2
1
Frequency GHz 10
2nd SKADS Workshop 10-11 October 2007
100
Andrew Faulkner
Conceptual SKA Configuration
Comms links
Core (5Km dia),
made up of close
packed stations
Central
Processing
in or near Core
Station
Not to scale!
Desert
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Station
Processing
Link to
Correlator
High freq. Dishes with
analogue fibre link
EoR AA antennas with
analogue link (may
be close packed)
Notes:
1. Power dist. not shown
2. All analogue links go to
‘station processing’
Station
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Station Processing
Bunker
High Freq.
dishes
1st
Stage Processors
.
Dish P1
Dish P2
.
.
.
..
Station
Processor 1
Dish Px
20GHz Analog fibre links
Station
Processor 2
.
Mid Freq.
AA
…..
Mid P1
Mid P2
Mid Py
Station
Processor X
0.3-1.0GHz Analog links
Digital
Systems
Internal
Digital links
.
.
Only
Low Freq.
AA Analogue
Low P1
Low P2
To Correlator
RFI
Barrier
n x Optical
fibres per 2nd
stage processor
.
..
.
Low Pz
300MHz Analog links
2nd SKADS Workshop 10-11 October 2007
Phase
Standard &
Distribution
Control
processors
Phase transfer
over fibre (where used)
To Central
Control system
10Gb Digital
fibre links
Andrew Faulkner
Antenna Array…
~60m
Tile
Support
Bunker
•
•
•
•
Contiguous array
Maintain from underneath
Cover with RF transparent sheet
Bunker in centre
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Detailed design
Rack Mounting
0.2mm pitch Euro connectors
ARJ45 Backplane 7 rows (5 sig + 2 power)
Notes:
1. Metalwork not shown
2. Mount 16 per rack
3. Probably PSUs in the middle and the
bottom to keep leads short.
4. PSU probably 4U
Components
Large Power
Connector
Cooling water
port
3.5”
2U
Bunker racking
Rack cabling
2mm pitch connectors
ARJ45 Backplane 7 rows (5 sig + 2 power)
Notes:
1. Mechanical and rack support not shown
2. 4 rows of 32 ARJ45 connectors give 512
differential inputs. Total 1024 connections.
3. Backplane connectors (50mm long) have
125 signal lines per block, plus 50 power
and ground pins.
4. Total signal lines through Euro connectors
is 1250.
5. Main board plugs into backplane carrier
and is individually connected for power and
output.
6. Backplane board permamently fixed into
rack
7. *16 x digital outputs used in digital system
(8 FoVs with 2 polarisations).
Components
Large Power
Connector
Cooling water
port
3.5”
2U
Coolant
Flow
Power
Time Standard
Control network
To Station
processors
~550mm
Overall Sysytem:
Spacing:
No. per rack:
No. of racks:
Connectors (ARJ45) :
Connector pitch
Total connector width:
Beams to
Station processors*
2U (3.5in / 88.9mm)
16 (total 32U)
16
4 rows of 32
0.6in
19.2in / 488mm
Design Blocks
Antenna array Note: water cooling
Analogue signal transport
Analogue beamforming
Digital Beamforming
2nd Stage Processing
Digital Data transport
Dishes
Mechanical Infrastructure
~500mm
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
First costing result
Total SKA cost breakdown
AA station
Correlator
7.9%
Calibration Source
0%
Signal
Processing
33%
Antennas
25%
Clock, control,
phase standard
1.4%
Mid Freq AA
Antennas
11.3%
Mid Freq AA
Analogue Data
Transport
9.7%
Comms.>480 km
9.6%
Comms.<480km
8.1%
2nd stage
processors
4.4%
Infrastructure
21%
Analogue Data
Transport
21%
Mid Freq AA Signal
Processing
15.0%
Common Station
Infrastructure
1.2%
Mid Freq AA
Calibration source
0.1%
Mid Freq AA
Infrastructure
9.5%
Total: €3.1M
Total: €1.9B
High Freq Dishes
19.2%
Low-Freq AA
2.5%
Projected in 2011 €
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Pros and Cons
PROS
• Design and engineering led
• Involved most engineering
SKADS groups
• Focussed thinking on issues
• A complete SKA
CONS
• Hard to vary the design:
– Mix of technologies
– Geometry
– Performance
• Technology based
• Light on ‘non-SKADS’ systems
– Central processing
– Dishes
• Connection to science
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Draft Specifications for the
Square Kilometre Array
The output from a ‘Specification Tiger Team’:
R. T. Schilizzi, P. Alexander, J. M. Cordes, P. E. Dewdney, R. D. Ekers,
A. J. Faulkner, B. M. Gaensler, P. J. Hall, J. L. Jonas, K. I. Kellermann
Covers science requirements and implementation trade-offs at fixed
Cost covering three collector technologies.
The principal input into
SKA2007
Manchester: 27-28 September
Available at: www.skatelescope.org
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Sparse AA –
element type 3
1000
Fully sampled AA
Element type 1
100
Aeff
No new elements –
maybe sparse
above fAA
10000
10
Aeff/Tsys
5000
f sky
1
100
Aeff / Tsys (m2 / K)
Sky Brightness Temperature (K)
Sparse AA –
element type 2
f AA
500
f max
1000
1000
3000
Frequency (MHz)
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
SKA Spec: Key points
• Describes options for SKA low and mid up to 10GHz
• Has a Baseline design with Wide FoV technologies ‘expectations’
• All designs use 15m dishes
– for risk mitigation: low frequency performance & computing cost
• Restricted high frequency operation to 2020, high freqs to follow
• Indicated performance for Key Science Projects
AA for ≤500 MHz in
All designs
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
TT proposals after SKA2007:
Baseline
Design
Focal Plane
Array WFoV
Aperture
Array WFoV
Sparse AA:
Sparse AA:
3100 x 15m dishes+WBSPF:
70-200 MHz
200-500 MHz
<500 MHz–10 GHz
Sparse AA:
Sparse AA:
1615 x 15m dishes+PAFs:
+WBSPF:
70-200 MHz
200-500 MHz
500 MHz–1.4 GHz
1.5 GHz – 10 GHz
Sparse AA:
Sparse AA:
Dense AA:
3100 x 15m dishes+WBSPFs:
70 – 200 MHz
200 – 500 MHz
500 – 1000 MHz
800 MHz – 10 GHz
There can be many variations!
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Evolving the Benchmark
Specification D&C-2
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
D & C-2: Goals
• Communicate SKADS D&C work Internationally
• Bring science experiment reqts into the design process
• Include new Design Blocks
• Update existing Design Blocks with new information
• Optimise System Configurations
• Use information from other international groups
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Some Challenges:
ID
Challenge
Consequences/ Mitigation
1.
Developing 4:1 frequency range antenna elements
2.
Processing capability within power budget
3.
Minimum Tinst – specification:
4.
Mechanical design and on-site construction for sparsed arrays.
System design must be completed first.
Increased cost
5.
Power requirements and cooling.
Increased running costs
6.
Self induced RFI within acceptable limits
7.
Multiplexing channels for AA-lo to reduce total processing
requirements
Increased costs
8.
Demonstration of systematics below level required for the
experiments
Unable to perform the science
9.
Maintainability and reliability within agreed limits
Increased costs
AA-lo of 100 K
AA-hi of 30 K
Use 3 different frequency bands
Reduced sensitivity or higher cost if not
met. Less critical for AA-lo
From: Draft Specifications for the Square Kilometre Array
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
New Design Blocks
Science drivers
Central processing
Array geometry
Low System Noise
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
New Design Blocks
Science drivers
4 hr
integration
Central processing
360 hr
integration
Array geometry
10-sigma
Aeff/Tsys = 20,000
Low System Noise
e.g detect M* galaxy at
z=0.75 in 25 hours
with Aeff/Tsys= 10,000
Abdalla and Rawlings
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
New Design Blocks
Science drivers
Potentially a very big issue
Central processing
May be the highest system cost!
Calculations by Tim Cornwell, see:
SKA Memos: 49 and 64
Array geometry
The AAs could:
Low System Noise
be good – low N large D
or bad – many beams, wide FoV
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
New Design Blocks
12,000
10,000
Total Sensitivity
AA-lo
AA-hi sparse
f AA
f sky
10,000
AA-hi close packed
1,000
/K
f max
sky
2
Aeff/Tsys m
Central processing
-K
Sky Noise
8,000
6,000
100
4,000
Dense?
Sky noise T
Science drivers
10
2,000
0
0
Array geometry
200
Total Sensitivity
1
25,000
600
400
800
AA-lo
AA-hi
1000
Frequency MHz
10,000
Sky Noise
20,000
sky
10,000
100
Sky noise T
2
Low System Noise
15,000
Aeff/Tsys m
/K
K
1,000
10
5,000
Sparse?
0
1
0
200
400
600
800
1000
Frequency MHz
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
New Design Blocks
• Critical cost driver: 1 K worth ~€10M
Science drivers
– or 2% more sensitivity
Central processing
– or 4% more survey speed
• Sky noise important at these freqs
Array geometry
– Minimise Trec then +Tsky = Tsys( f )
Low System Noise
• All contributions must be investigated
Key demonstration to the
International community
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner
Programme Timetable
Initial D&C - 2 meeting
13 September 2007
Work on science reqts & system blocks
Review Meeting
31 October 2007
Preparing written work + science & technical
details
Initial Drafts received
Review Paper draft (telecon)
9 November
23 November
Consolidating paper
Final Draft
7 December
Approval of Paper
11 December
Publication of Paper
12 December
SKA Specification Review Committee: Jan 2008
2nd SKADS Workshop 10-11 October 2007
Andrew Faulkner