1 of 32

ngVLA Project Summary

Eric J. Murphy – Project Scientist

Italian Radio Astronomy School: �2 October 2025

2 of 32

Big Picture - Overview

The ngVLA will be a single interferometric array that replaces the NSF Jansky Very Large Array and the NSF Very Long Baseline Array.

Astro2020 identified the ngVLA as a high-priority, ground-based large facility whose construction should start this decade.

2

ngVLA concept

Frequency span 1.2 – 116 GHz
Resolution span 0.1 milliarcsec – 10 arcsec
10 x sensitivity of the Jansky VLA and ALMA
244 x 18m + 19 x 6m offset Gregorian antennas

At fixed locations in the U.S. and Mexico
Concentrated in the U.S. Southwest

ngVLA science bridges SKA/ALMA

SKA

ngVLA

ALMA

Atacama Large

Millimeter/submillimeter Array

Square

Kilometer Array

ngvla.nrao.edu

3 of 32

Terrestrial zone planet formation: 1AU @ 140pc

Linking SKA & ALMA Scientifically

Complementary suite from cm to submm arrays for the mid-21^st century

< 0.3cm: ALMA 2030 superb for chemistry, dust, fine structure lines

0.3 to 3cm: ngVLA superb for terrestrial planet formation, dense gas history, baryon cycling
> 3cm: SKA superb for pulsars, reionization, HI + continuum surveys

ngVLA

Sensitivity

Resolution

4 of 32

ngVLA Project

8 Integrated Product Teams (IPTs).
MREFC-style project definition.
Actively engaged science/technical �advisory councils.

4

5 of 32

5

Executive Committee

Brenda Matthews (NRC–Vic)

David Wilner (CfA/SAO: co-Chair/SWG1)

Carlos Carasco-Gonzalez (UNAM: SWG1)

Brett McGuire (MIT: SWG2)

Fabian Walter (MPIA: SWG3)

Alexander van der Horst (GWU: SWG4)

Rachel Osten (STScI: SWG5)

Alberto Bolatto (Maryland: ex-officio)

Andrea Isella (Rice: ex-officio)

Interface between the Community & NRAO -- Est. Sept 2016
Recent/Current Activities:

Lead SWGs: science use cases 🡪 telescope requirements
SOC participation for ngVLA science meetings
Lead Science case development 🡪 ‘Science Book’ & Astro2020 White Papers
Document Review: e.g., Sci Reqs, Ops Con, Ref Observing Program, etc.
Updated KSG Memo (2024: Astro2020/ngVLA Science Gap Analysis)
ngVLA Critical Capability Derivation

ngVLA

Science Advisory Council

International Participation

6 of 32

ngVLA Community

Science Advisory Council offers expertise, guidance and feedback, and leads Science Working Groups with 300+ subscribers
Technical Advisory Council offers expertise, guidance and feedback on engineering and computing topics
Sought use cases, Science Book chapters, white papers
Supported 50+ Community Studies
Showcased 50+ scientific papers in NRAO eNews
Supported 30+ scientific and technical conferences
ngVLA mentioned in 1200+ community publications

6

Proactively engaged the worldwide scientific and technical communities since 2015

Credit: Brian Kent

ngvla.nrao.edu

7 of 32

NSF RIG: MREFC Design Queue �The Conventional Path

7

From MPS 2023 Major Facility Report

Nominally

Funded

Technical Baseline

July 2022

NSF CDR �September 2024

NSF PDR

2026?

Identified as a formal MREFC Design Candidate by NSF

Awarded ($21M) over 3 years (FY24-FY26) to Support PDR

HERE ☹

8 of 32

ngvla.nrao.edu

Unveiling the Formation of Solar System Analogues on Terrestrial Scales

Probing the Initial Conditions for Planetary Systems and Life with Astrochemistry

Charting the Assembly, Structure, and Evolution of Galaxies Over Cosmic Time

Science at the Extremes: Pulsars as Laboratories for Fundamental Physics

Understanding the Formation and Evolution of Stellar and Supermassive �Black holes in the Era of Multi-Messenger Astronomy

ngVLA Key Science Goals�(ngVLA memos #19 & 125)

Science requirements

Technical concept

9 of 32

9

KSG2

KSG3

KSG4

KSG5

KSG1

10 of 32

10

ngVLA Reference Design

Band�#	Freq. Range (GHz)
1	1.2 - 3.5
2	3.5 - 12.3
3	12.3 - 20.5
4	20.5 - 34
5	30.5 - 50.5
6	70 - 116

Frequency Range: 1.2 - 116 GHz
Main Array: 244 x 18m offset Gregorian Antennas

Core: 114 antennas; B_max = 4.3 km
Spiral: 54 antennas; B_max = 39 km
Mid: 46 antennas in NM, AZ, TX, MX; B_max=1070 km
Long: 30 antennas across continent; B_max= 8860 km

Short Baseline Array: 19 x 6m offset Greg. Antennas

Use 4 x 18m in Total Power mode to fill (u,v) hole

Technical Concept

Key design choice: Antennas in fixed locations

Year-round access to all angular resolutions
PI-driven facility providing science subarrays

Correlator / Beamformer	Requirement (design)
digital efficiency	>95%
narrowest channel	<1 kHz
total # channels	>240,000
sub-band width	<250MHz (218.75)
total bandwidth	>14GHz/pol (20)
# formed beams	10

11 of 32

11

Re-Baselined�Configuration

12 of 32

S/W and Computing Considerations

Operations Concept: HLDP (High-Level Data Product) Telescope

Both for 1^st Observations and Archive projects.

Post Processing: Analysis shows that storing the raw visibilities will be tractable when ngVLA goes into operations.

Data processing is post-facto, with system sized for average throughput.
Average Data Rate – 7.6 GB/s. Designed for 320 GB/s peak.
4 hr. observation – 109 TB. Requires ~1000 cores to process in a few days.

Computing: 2B Core-hr: Challenging, but can be met w/ COTS cluster.

Set by time resolution, spectral resolution, and multi-faceting in imaging
Some low-frequency, full-beam, AW-projection cases restricted in early operations.

12

Science-Ready is a loaded phrase… point is that data projects can be used for scientific analysis

Max rate: 320 Gbps FE -> CSP

7 TeraVis/hour = 7.6 GB/s (Average) -> 109 TB in 4 hr

120 TeraVis/hour = 130 GB/s (peak) -> 1.9 PB in 4 hr

1500 standard (without W/A-projection)

Can grow up by ~x1000

** 60 PFLOPs/s --- size of the system (factors efficiencies – size of Frontera UT super computer 40 PFLOPs)

We are big… Moore’s law has flattened… so improvement from different areas.

For Spectra: Parallelization very efficient (Embarrassingly parallel l!!)

Most cases, require lots of channels (>100k) -- each channel can be processed independently.

Challenge is just processing a lot data, to achievable.

External Processing centers (e.g., OpenScienceGrid)

Accommodate projects whoe throughput beyond our capabilities
E.g., 1 big survey project exceeds 60 PFLOP/s, exceeds latency needed for data delivery

OpenScienceGrid (High Throughput Computing) – golfballs in container vs. basketballs – increase granularity to accommodate flexibility.

Group of universities to join computing with common software (i.e., different than EXCEED)
Part of LHC – largest computing problem in public sector (1B core-hour core problem… ngVLA will be 2B)
HTC framework dividing jobs into small junks sent to different facilties.

13 of 32

Recent Highlights

Completed NSF-run CDR September 3 – 6, 2024

Panel recommended project move to PDR & Design be fully funded

Supported a 3 AAS Splinter Sessions and Focus Meeting at IAU GA
Held 2024 ngVLA Science Meeting in MX

Morelia; broad range of science using ~100+km baselines
Increased in engagement with multiple Universities/States in MX

ngVLA Science Advisory Council released updated Key Science Goals

Identifies expected data product and computing needs

Completed Community-Led VLA/VLBA🡪 ngVLA Transition Options Concept
MOUs signed with:

University of Florida, Johns Hopkins, and Texas Tech

3 New Science Hires (2 Sci. Staff & 1 Research Associate)
Antenna Prototype Handover Event April 25^th

13

14 of 32

FY25 Goals: System Design Phase

Continue to build and hone the ngVLA science case with community

Update/incorporate additional science cases for system design

Complete open work/efforts that will support the system PDR

Update and synchronize technical and programmatic documents
Large effort to allow software to catch up: CDR this Fall
Continue antenna development, including USNO/Low Elevation variant

Secure domestic and international partnership contributions for final design and construction.

Baseline plan and work package distribution by PDR.

Acquire additional funding that will help steer project towards a Final Design

Early Science ~2032 and Full Science mid/late next decade.

14

15 of 32

International Engagement

International involvement via SAC, TAC, Community Studies

Canada, Mexico, Japan, Germany, Netherlands, Taiwan

NAOJ-ngVLA workshop, Mitaka (2019) – 1^stinternational science meeting
ngVLA currently a candidate being considered as part of Japanese Master Plan
ngVLA identified as a high priority future project in Canadian LRP (6% partner)
Signed MOU with UNAM (11/4/22) for their participation in ngVLA design.
German ngVLA workshop held in Bonn (12/14/22) – ~100 attendees!
Science Meeting + mtex Open House held in Leipzig (9/23).
Mexican-led ngVLA Science Meeting in Morelia (9/23 & 11/24)
MOA signing between Sonora, Chihuahua, UNAM, INAOE (CDMX 9/25)
MX visit with UNAM Rector, SECIHTI, US Embassy, (9/25)

16 of 32

Long Baseline Antenna Development (USNO)

FY2024-FY2025 Award developing a Long Baseline Antenna design variant incorporating revised USNO requirements. [~$1.6M]
Deliverables:

Technical Requirements for Long Baseline Antenna (Delivered)

Doc: 020.28.00.00.00-0001 REQ.

Prototype Site & Test Plan Evaluation (Delivered)

ngVLA Antenna Memo #14, Long Baseline Antenna Prototype: Test Site Selection
ngVLA Antenna Memo #15, Long Baseline Antenna Prototype: Antenna Qualification Electronics Concept
ngVLA Antenna Memo #16, Long Baseline Antenna Prototype: DBE Concept

Project Plan for prototype construction and evaluation (Delivered)

NRAO Proposal #884

Preliminary Design Review (Passed – minor data pack updates pending)

But wait, there’s more…

16

17 of 32

Long Baseline Antenna Development

17

Key Differences to the Main Array Antenna:

7-deg vs 12-deg Elevation limit
Security protocol implementation
Slew velocity
Design iteration for performance and manufacturability.

Main Array Antenna (MAA)

Long Baseline Antenna (LBA)

18 of 32

ngVLA High Resolution Subarray (HRS):�NSF & USNO Supported Start of ngVLA

Possible GBO Antenna Location

18

Initial VLBA replacement
Low Elevation (USNO) Antennas
Combination of ngVLA and USNO electronics
~10 Sites: Mixture of existing and new sites, including GBO
Split USNO & PI Science Operations

Similar to VLBA now

Can be incorporated into ngVLA Long

Not simply a work for others project!

19 of 32

ngVLA High Resolution Subarray (HRS):�NSF & USNO Supported Start of ngVLA

19

Cost: ~$300M
USNO wants to continue NSF partnership & support science ops
NSF views this as continuation of ngVLA design towards construction
Timeline

HRS FDR FY27 (= ngVLA PDR)
Construction FY28 – 29
Early Ops FY29 🡪 (ngVLA FDR?)

Illustrative Example of Configuration

20 of 32

FY26 Plans

20

Advance ngVLA Technology that additionally supports GPA

Focus on getting pieces of ngVLA system PDR-ready in FY26
Mature software architecture/operations processes
Further define international partnerships
Engage (and grow) the US Long Baseline community

Continue with Antenna Development for Low Elevation Design

Start Scientific Testing of Prototype

January

ngVLA Science Meetings

Oct. 2025: Astrochemistry in �Portland ME
2026: Multidisciplinary Tohoku U., Japan
2026: Long Baseline Science in DC Area

21 of 32

Summary

The ngVLA continues to make progress and identify opportunities to work towards a completed design.

The next few years should remain busy:

Scientifically validate the prototype.

Prepare project for successful System PDR when possible

Button up open efforts that will have a long shelf life
Begin prioritizing ngVLA design work that additionally supports GPA design and construction effort.
Never lose focus of the ultimate goal

Continue to engage international and interagency communities to solidify partnerships

All made possible by strong community and NSF support.

21

22 of 32

ngvla.nrao.edu

23 of 32

KSG1: Unveiling the Formation of Solar System Analogues�

23

The ngVLA will measure the planet IMF down to ~5-10 Earth masses and unveil the formation of planetary systems similar to our own Solar System.

24 of 32

The ngVLA can detect complex pre-biotic molecules and provide the chemical initial conditions in forming solar systems and individual planets

24

KSG2: Probing the Initial Conditions for Planetary Systems and Life with Astrochemistry

Credit: Brett McGuire (NRAO)

This is a simulation of ~3 complex organic molecules, not currently detected in the ISM, but which are good candidates for detection. �

The simulation assumes that the emission is coming from a compact, 1" hot core in Sgr B2(N), from molecules with column densities of 10^12 - 10^14 cm-2, at T = 200 K, and with a linewidth of 3 km/s.

The simulation does *not* contain any currently-detected species, meaning that a real spectrum in this frequency range would have many, many bright lines off the scale of this graph.

The red horizontal line is the ~rms of the current GBT PRIMOS observations. These suffer greatly from beam dilution, as the 1" source is diluted in a 30" GBT beam. This rms is about what we can reasonably get with reasonable integration times at the GBT. Drilling deeper will take far longer than is feasible for a survey.

The green line is taken from the ngVLA tech specs page for line sensitivity at 10 km/s in 1 hr, degraded for 0.3 km/s channels.

The blue shaded region is the area in which I project, from a statistical analysis of the PRIMOS observations, we will hit line-confusion in this frequency range. There is little point in integrating more than 3 sigma deeper than this.

�

So what are the take-away messages here?

�

1) the ngVLA will almost certainly provide access to a wealth of new molecular detections of molecules similar to the representative species shown here

�

2) the ngVLA in reasonable integration times (say 4-20 hours) should hit line confusion in this most molecularly-rich galactic source at the 3 sigma level, something that the GBT and (J/E)VLA cannot hope to accomplish.

�

25 of 32

25

KSG3: Charting the Assembly, Structure, and Evolution of Galaxies from the First Billions Years to the Present

Credit: Caitlin Casey (UT Austin)

SMG at z = 4.4; SFR ≈ 400 M_⨀ /yr

Total molecular gas content largely missed by high-J lines

𝜃 ≈ 0.3”

𝜃 ≈ 0.5”

The ngVLA will routinely detect molecular gas in “normal” galaxies at z=6 via low-J transitions that remain inaccessible to ALMA.

The ngVLA will provide >10x improvement in our knowledge of the cold molecular gas content throughout cosmic time.

26 of 32

The ngVLA sensitivity and frequency coverage will probe deeper than currently possible into the GC area looking for pulsars, which are moving clocks in the space-time potential of Sgr A*

New tests of theories of gravity, constraints on exotic binaries, SF history, stellar dynamics and evolution, and ISM at the GC

Estimates are as high as 1,000 PSRs. Only known example is PSR J1745-2900 magnetar, which are extremely rare (<1%)

26

KSG4: Science at the Extremes: Pulsars as Laboratories for Fundamental Physics

Credit: R. Wharton

27 of 32

KSG5: Understanding the Formation and Evolution of Black Holes in the Era of Multi-Messenger Astronomy�

27

The ngVLA’s sensitivity & angular resolution will be able to:

Localize & Resolve dual AGN and BH binaries directly in the Radio.

Detect GW170817 source at Adv LIGO horizon dist. of 200 Mpc.

Measure proper motion expansion over 5 year periods (orange shaded region), including GW sources

Search for BHs across all masses

e.g., weakly accreting MW BHs & SMBHs in nearby dwarfs via proper motions
Increase sample by ~x10

28 of 32

Science Highlight: Star Formation and Stellar Evolution

28

The ngVLA will measure the in –situ gas motions from material shed around AGB stars.

Credit: K. Akiyama & L. Matthews

based on models from B. Freytag

Supported by ngVLA Comm Study Program

Simulation based on 3D hydrodynamic model of AGB star Atmosphere from Freytag et al. (2017):

ngVLA Main Array at 46 GHz
1.5 mas ~ 0.04 stellar radii at d=150pc
1.3 year pulse period
Observed every 2-3 weeks

ngVLA Memo #66

Simulated movies of the gas motions in the radio photosphere of an asymptotic

giant branch (AGB) star as observed with the ngVLA Main Array at 46 GHz. The

angular resolution is ~1.5 mas, corresponding to roughly 0.04 stellar radii.

Each movie covers a single 1.3-year pulsation period for a model AGB star,

assuming observations every 2-3 weeks. The adopted distance was 150 pc.

AGB star atmospheres are highly dynamic owing to a combination of radial

pulsations, shocks, and large-scale convective processes. These stars also shed

large amounts of matter through cool, dense winds. Radio waves are emitted from

a portion of their atmospheres that lie at a radius of roughly twice the

visible photosphere, and just interior to the wind launch region.

The adopted brightness distribution for the radio photosphere model (the

"ground truth model" shown in the left hand frame) was based on a 3D

hydrodynamic model of an AGB star atmosphere from Freytag et al. (2017). The

Freytag et al. model is not specific to millimeter wavelengths, but was adapted

here for illustrative purposes. The second frame shows the ground truth model

blurred with a Gaussian kernel.

The two frames on the right show image reconstructions of the model based

on simulated ngVLA Main Array observations. The integration time was 2

hours per frame and the assumed bandwidth was 10 GHz in dual

polarizations. The second frame from the right shows a multi-scale CLEAN

reconstruction, while the frame on the right shows the result of a

regularized maximum likelihood reconstruction using the SMILI package of

Akiyama et al. (2019). While both methods do well at reproducing many of

the key features of the model, the SMILI version achieves superior angular

resolution.

These results underscore that the exquisite sensitivity and resolution of the

ngVLA will enable studies of the physical properties and kinematics of the

atmospheres of evolved stars in unprecedented detail.

Credit: K. Akiyama and L. D. Matthews, based on models adapted from B.

Freytag.

29 of 32

29

Re-Baselined�Configuration

30 of 32

Antenna Electronics Development�(Front Ends)

Band 1 (EMMS/NRAO):

Work has just begun and development is in-kind.
Will provide drawings, with the intention of submitting a quote for the project proposal.
Callisto will be supplying the coldhead.

Band 2 (CSIRO):

CSIRO is nearing completion of its involvement.
Have conducted feed horn pattern testing; final report expected soon.

Band 3: No updates provided.
Band 4: No ongoing activity.
Band 5 (NRC/NRAO):

NRC & NRAO signed MOU.
NRC has just begun work on the vacuum window.

Band 6 (NAOJ):

NAOJ has expressed interest, though no formal agreements have been signed yet.

30

Band 2

31 of 32

Antenna Electronics Development�(Cryogenics)

A two-stage Gifford-McMahon cryogenic system is being utilized

variable-speed cryocoolers and compressors operating at 40-90 rpm.

Scroll-type helium compressor (Hitachi) installed in antenna turn-head.

It runs on three-phase 408 VAC power and is liquid-cooled.
Due to its inability to tilt, it must be positioned below the elevation axis, inside the turn-head, where it is shielded from weather exposure.

A helium backup is also located within the turn-head.

31

32 of 32

Main Antenna Dev: Germany

Wetterstein Millimeter Telescope (WMT)

Atmospheric observatory on highest mountain in Germany

LEVERAGE: Long-baseline Extension in next-generation VLBI Experiments and Rapid-response Array GErmany

32